Biology, Behaviour and Taxonomy of two Oleria onega ... · Chapter 4 Modelling of the Distribution...

Université de Neuchâtel Institut de Zoologie _________________________________________________

Biology, Behaviour and Taxonomy of two Oleria

onega subspecies (Ithomiinae, Nymphalidae,

Lepidoptera) in north-eastern Peru

Stéphanie Astrid Gallusser

Laboratoire d’Ecologie Animale et Entomologie

Thèse présentée à la Faculté des Sciences de l’Université de Neuchâtel pour

l’obtention du grade de docteur ès science

2002

CONTENTS

Thesis abstract.................................................................................................................…..................1 Résumé...................................................................................................................................................4 General Introduction........................................................................................................….……..........7 Thesis outline......................................................................................................................................11 Chapter 1. Host-plant preference and oviposition behaviour of two Oleria onega subspecies

(Ithomiinae Lepidoptera) in north-eastern Peru.............................................................16 Chapter 2. Comparison of larval behaviour of two Oleria onega subspecies (Ithomiinae,

Lepidoptera), in relation to predators of their early stages............................…...….....42 Chapter 3 Genetic (RAPD) diversity within and between Oleria onega agarista and Oleria onega

ssp. (Ithomiinae, Nymphalidae, Lepidoptera) in north-eastern Peru............................67 Chapter 4 Modelling of the Distribution of Oleria onega agarista and Oleria onega spp.

(Ithomiinae, Lepidoptera), in Peru Using Geographic Information Systems and Geospatial Analysis............................................................……………………..................86

Annexes…………………...............................................................................…...................................106 Conclusion and outlook....................................................................................................................109 Curriculum vitae...........................................................................…...................................................113

Thesis abstract - Résumé

THESIS ABSTRACT

Oleria onega agarista (Felder & Felder) and Oleria onega ssp. nov. are Ithomiinae subspecies

that live on the Upper Huallaga River, near the town of Tarapoto in north-eastern Peru, and are

separated by a mountain chain, the Cerro Escalera. Oleria onega agarista live on the NE slope and O.

o. spp on the SW slope. We conducted both observational studies and field experiments to describe the

ecology of these two taxa and to make hypotheses on potential selection pressures acting in the

evolution of the two taxa.

1. The two Oleria subspecies are morphologically different and we document in the first part of

this study different oviposition behaviour. Host plant preferences between four Solanum species, S.

mite, S. anceps. S. angustialatum and S. uleanum, were tested and revealed that Solanum mite is the

most preferred host plant by the females of both butterflies subspecies, and is also the most abundant,

followed by S. anceps, uleanum and angustialatum. Oleria onega agarista laid their eggs on its host-

plant, Solanum mite, whereas Oleria onega ssp. laid them at a distance, till up to 1 m. away from the

plant. Through experiments in cages and observation in the field, it appears that this behaviour is not

always the case, and that both Ithomiinae subspecies may present both behaviour, but in the field O. o.

ssp. prefer to oviposit away from the plant, a new strategy for this genus. The main selection pressure

that could lead to the oviposition behaviour differences could to be the predation on eggs and larvae,

increased by competition for oviposition site between ovipositing females. The higher density of host-

plant on the SW slope, allowing the larvae to find their food easier, could have helped the development

of this alternative strategy.

2. Four Solanum species were offered to the larvae of both Oleria species. The results revealed

that Solanum mite is the most preferred host by both females and immature stages. Preferences of

ovipositing females and larvae are highly correlated with the performance of the larvae on S. mite. Egg

survival in the field was tested for both Oleria subspecies on both side of the mountain. For this, a total

of 400 eggs were glued, half of them on S. mite leaves and half on other substrates. No differences

were found between the two butterfly subspecies, however egg survival on the SW side was higher

when glued on other substrates than when glued on the host plant leaves. The O. o. ssp. larvae (natural

of the SW side) use to move less than those of O. o. agarista do. Therefore a hypothesis for the

behaviour of laying eggs away from the host plant was a higher predation pressure on the SW side.

1


Predator fauna was compared between both mountainsides, revealing that 70-80% of the potential

predators were ants. Among them, Ectatomma sp. (Hymenoptera: Ponerinae) was found in high

numbers on the SW side, but were completely absent of the NE side, where no predation events were

observed on eggs. Therefore we emit two hypotheses: first Ectatomma sp. may eat or remove the eggs

from the plant, or their presence on plants may avoid oviposition by the females.

3. Two contact zones on the SW slope are known: Ahuashiyacu, where both subspecies only

cohabit, and Estero (near the village of Shapaja), where they hybridise. Genetic differences between

the two subspecies and between populations were investigated with Random Amplified Polymorphic

DNA (RAPD) markers. Both cluster and Principal Coordinates Analyses provided a clear but weak

discrimination between the two subspecies. Genetic diversity was much higher within the populations

than between them, and following the results of the cluster analysis, the geographically more distant

populations were grouped. Morphological traits on the wing patterns of the hybrids are intermediary

between the two butterfly subspecies, but it should be noticed that hybrids showed fewer common

RAPD bands with the two subspecies than expected. Nevertheless, in the analysis they presented more

similarities with O. o. agarista, from the Km 28 population, and only with three O. o. ssp. individuals

from the Urahuasha populations that may be back crosses from hybrids. The individuals of the

Ahuashiyacu population were well defined between O. o. ssp. and O. o. agarista, according to

morphological traits, and no individual was near the hybrids, suggesting that hybridisation has not yet

occurred in this population, even though both subspecies occur sympatric. As polymorphism is high

(100%), the use of specific markers and analysis of other subspecies of the Oleria onega complex will

help to determine the exact taxonomical status of Oleria onega ssp. and O. o. agarista and their

hybrids. Even so, RAPD techniques provided good information for discriminating between the

subspecies and among populations.

4. We attempt to model the distribution of the butterflies in relation to environmental variation,

using GIS and geostatistical tools; and this part is divided in four main steps: (i) through punctual

observations of twenty-three sites, cartographic maps on presence and absence of butterflies, the

presence of the different host plant species, the type of soil, dimensions of trees and the density of the

understorey were built. (ii) Thereafter temperature, humidity and density of the butterflies were

measured for each sites. However, because of the lack of data from part of the area, synthetic data

needed to be generated. Through statistical and interpolation methods, both the spatial distribution of

2


Oleria onega subspecies and environmental data were extrapolated to have data available for all the

area. Topography and landuse were represented for all the area. (iii) Conditional simulations allowed

the estimation of probabilities of presence of 1, 2, 3,4 and 5 individuals in all the study area, revealing

that the probability of finding two individuals/day is closest to the actual situation. (iv) Kriging,

probabilities, landuse, topography and buffers of the rivers, study sites and road were compared. The

mean number of both males and females observed per day was between 0.8-1.5 individuals in most of

the area. This mean number is situated mostly between 85-90% of humidity and 25-27° C. In relation to

altitude, butterflies were found between 300-800m and in forested areas. In two areas, the abundance

of butterflies was higher, where pathways through the mountain are possible.

3


RESUME

Oleria onega agarista (Felder &Felder) et Oleria onega ssp. nov. sont deux sous-espèces

d'Ithomiinae qui vivent dans la région du Haut-Huallaga, près de la ville de Tarapoto, dans le nord-est

du Pérou. Elles sont séparées par la chaîne de montagne appelée "Cerro Escalera" : Oleria onega

agarista se trouve sur le versant nord-est (NE) et O. o. spp. sur le versant sud-ouest (SO).

1. Elles présentent des différences morphologiques et des différences dans le comportement

d'oviposition. Dans la première partie de cette thèse nous avons testé leurs préférences pour quatre

espèces de Solanum, considérées comme plantes hôtes. Solanum mite, qui est l'espèce la plus

abondante, s'est révélée être la plante hôte préférée par les femelles des deux sous-espèces suivie de

Solanum anceps, S. uleanum et S. angustialatum. Oleria onega agarista dépose ses oeufs sur les

feuilles de la plante hôte alors que Oleria onega ssp. les dépose plutôt sur des objets avoisinants,

jusqu'à 1 mètre de distance de la plante hôte. À l'aide d'expériences en laboratoire et d'observations sur

le terrain, ce comportement s'est révélé être fréquent chez O. o. ssp., surtout sur le terrain, bien que les

deux sous-espèces d'Oleria en cage puissent parfois présenter les deux comportements. Le fait de

déposer les oeufs loin de la plante hôte est une stratégie de ponte nouvelle, encore inconnue du genre

Oleria. La principale pression sélective qui pourrait induire ce comportement nouveau est la présence,

sur le versant SO de la montagne, de prédateurs des œufs ou des femelles entrain de pondre. Un

facteur pouvant favoriser le développement de ce comportement est une densité plus grande de

Solanum mite sur le versant SO, faciliterait ainsi la recherche de nourriture pour les jeunes larves.

2. Les quatre espèces de Solanum ont été présentées aux larves provenant des deux sous-

espèces d'Oleria. Les résultats de cette expérience ont montré que, à l'instar des femelles, Solanum

mite est aussi la plante qu'elles préfèrent. De ce fait, les préférences des femelles et de leur progéniture

sont fortement corrélées entre elles, mais aussi avec les performances des larves sur S. mite.

La survie des oeufs sur le terrain a été testée sur les deux versants de la montagne avec les

deux sous-espèces d'Oleria. Pour cela, nous avons placé 400 oeufs sur le terrain (200 de chaque côté

de la montagne), dont la moitié ont été collés sur les feuilles de Solanum tandis que l'autre moitié ont

été collés sur d'autres substrats, proches des plantes hôtes. La survie des oeufs collés sur d'autres

substrats était bien meilleure sur le versant SO, que celle des oeufs collés sur les feuilles des plantes

hôtes. Sur le versant NE, la survie était à peu près égale sur la plante et sur d'autres substrats. Aucune

4


différence significative a été trouvée entre la survie des œufs des deux sous-espèces. D'autre part, les

larves d'O. o. ssp (qui proviennent du versant SO) sont plus sédentaires que celle d'O. o. agarista. Ces

résultats laissent supposer que la pression due aux prédateurs est plus forte sur le versant SO et peut

induire le comportement d’oviposition particulier d’O. o. ssp. Un relevé de l’épifaune prédatrice réalisé

sur les deux versants de la montagne, a montré que le 70 à 80 % des prédateurs potentiels capturés

étaient des fourmis. Parmi elles, un genre, Ectatomma (Hymenoptera: Ponerinae), a été trouvé en

abondance sur le versant SO, mais était complètement absent du versant NE. Malheureusement,

aucun cas de prédation sur les oeufs par ces fourmis n'a pu être démontré. Par contre nous pouvons

supposer que leur présence peut affecter les papillons, soit par des actes de prédation sur les oeufs,

soit en empêchant les femelles de pondre sur leur plante hôte, ce qui demande plus d'observations de

terrain.

3. Bien que les deux sous-espèces de papillons soient chacune d'un côté de la montagne, deux

zones de contact entre elles ont été observées sur le versant SO. Les deux sous-espèces cohabitent

Ahuashiyacu , et des hybrides ont été observés a Estero. Les différences génétiques entre les deux

sous-espèces et leurs hybrides, et les variations entre populations, ont été étudiées en utilisant la

technique du Random Amplified Polymorphic DNA (RAPD). Les résultats, présentés sous forme d'arbre

non-enraciné (UPGMA cluster), et d' Analyse des Coordonnées Principales (Pcoord), ont tous les deux

démontré que les deux sous-espèces sont bien distinctes. Les hybrides, bien que morphologiquement

intermédiaires, ont peu de bandes en commun avec les deux sous-espèces. Ils sont situés entre les

deux sous-espèces dans l'analyse des coordonnées principales, bien que mélangés avec la population

d'O. o. agarista du Km28 et quelques individus de celle de Urahuasha (O. o. ssp.). Ces deux

populations pourraient de ce fait présenter des signes d'hybridation passée. La diversité génétique est

aussi plus grande au sein des populations que entre populations de la même sous-espèce, et des

populations géographiquement éloignées sont génétiquement proches. Le polymorphisme des

marqueurs génétiques neutres étant extrêmement élevé pour toutes les populations, il serait intéressant

d'utiliser d'autres marqueurs spécifiques (isozymes, RFLP) et d'introduire d'autres espèces d'Oleria

dans des prochaines analyses, ce qui nous permettrait de mieux déterminer le statut taxonomique des

deux sous-espèces. Néanmoins la technique des RAPD nous a donné d’importantes informations sur la

distinction entre sous-espèces, sur les similarités entre populations, et sur les variations génétiques au

sein des populations.

5


4. Nous avons essayé de modéliser la répartition géographique des deux sous-espèces

d'Oleria avec les données environnementales les concernant (température, humidité, topographie etc.)

en utilisant les outils du Système d'Information Géographique (SIG) et de l'Analyse Géospatiale : (i) au

moyen d'observations ponctuelles sur 23 sites, des cartes de distribution des papillons, de leurs plantes

hôtes, du type de sol, de la taille des arbres et de la densité de végétation du sous-bois ont été créées.

(ii) Les moyennes de température, humidité et densité des papillons ont été mesurées dans chaque

site ; le manque de données dans certains endroits nous a poussé à injecter des données synthétiques

(estimées d'après la connaissance du terrain). À l'aide de méthodes de statistiques et d'interpolation

des cartes de krigeage de la densité des papillons, de la température et de l'humidité relative ont été

construites sur toute la zone d'étude. Le "landuse" et la topographie ont été représentés sur toute la

zone d'étude. (iii) Au moyen de simulations conditionnelles, des cartes de probabilités de présence des

papillons ont été établies pour 1, 2, 3, 4 et 5 individus par jour sur toute la zone d'étude ; la probabilité

de trouver deux ind. par jour s'est avéré être celle la plus proche de la réalité. (iv) les cartes de

krigeage, probabilités, "landuse", topographie de même que les "buffers" construits sur les rivières, la

route et les sites d'études, ont été comparés visuellement. De cette étude, nous pouvons conclure que :

l'abondance des mâles et des femelles varie entre 0.8-1.5 ind./jour et est la plus répandue sur toute la

zone d'étude. Cette densité se retrouve surtout entre 85-90 % d'humidité relative et 25-27° C. Les

papillons se trouvent surtout dans une tranche d'altitude entre 300-800m, dans les zones forestières.

Deux zones montrent une abondance plus grande de papillons. Suivant les données d'humidité, de

température et de topographie, le passage des papillons est possible entre les versants, en contournant

la montagne.

6

General Introduction

GENERAL INTRODUCTION

GENERALITIES ON ITHOMIINAE

The Ithomiinae are a subfamily of the Nymphalidae that include eight tribes endemic to the tropics

of Central and South America : Tithoreini, Melinaeini, Mechanitini, Napeogenini, Ithomiini, Oleriini,

Dircennini, Godyridini. A ninth tribe, the Tellervini, includes only one polytypic genus found in New

Guinea, Celebes and Queensland (Brown, 1987b). A total of 40 genus and more than 400 species have

been described. However their systematics remains unclear and very complex due to the high

frequency of mimicry in the genus and to geographic variation in colour pattern. We lack molecular

phylogenies for this group. Ithomiinae are medium sized butterflies, with wingspans varying between 2

and 10 cm. Traits of wing venation are the principal taxonomical characters (Lamas, in prep). The wings

are either brightly coloured with brown, orange, yellow and red, or transparent with black veins and

bearing some tiny hairs and no scales in the transparent parts. Males possess one or two androconial

hair patches or scent scales on the dorsal costal margin of the hindwings. When males are not looking

for females, their hair patches remain flat, covered by the forewing margin.

Ithomiinae prefer shady habitats and undisturbed forests. During the dry season they gather in

multispecies groups with high densities of individuals (25 ind/m2 ) (Brown & Benson, 1974; Brown,

1979; Vasconcellos-Neto, 1980). These gatherings are called "Ithomiinae pockets". Ithomiinae have a

high dispersal ability and unlike their sister group Heliconiinae (Brown & Vasconcellos Neto, 1976;

Mallet, 1986) lack a definite "home range". Their mobility and their relative longevity (up to 6 months)

allow a rapid gene flow between subpopulations.

The interest in Ithomiinae began in the 19th century and was due to two main reasons. First they

were considered as prime distasteful models in mimicry complexes throughout the Neotropics (Bates,

1862; Müller, 1878, 1879; Brown, 1974b). More recently, because they feed on host-plants that are rich

in alkaloids and other poisonous substances they represent a classical case of insect-plant biochemical

coevolution (Brown et al., 1991). In the 1970's, work on biochemical particularities of Ithomiinae and

Danainae were published, revealing complex insect-plant interactions (Edgar & Culvenor, 1974;

Boppré, 1978; Edgar 1982) as it was discovered that pyrrolizidine alkaloids (PA's), the poisonous

compounds of adult butterflies, were obtained by seeking them in flower nectar and plant exsudates, but

7


not from the host-plants. Primitive Ithomiinae feed on Asclepiadaceae and Apocynaceae, which contain

cardiac glycosides, which are used as defences by adults. For example, Tithorea harmonia and Tellervo

zoilus, which feed on Apocynaceae, are the only species that incorporate alkaloids from the larval

foodplant (Prestonia acutifolia). The larvae of other species feed on Solanaceae which lack PA's

(Brown, 1984a,b) but contain other alkaloids, steroidal bitter principles, saponins and coumarins that

may be used by larvae as defences, but never by the adults (Brown, 1987a). Therefore adults need to

search for other sources of PA's in other plant species. Usually males are more attracted than are

females to flowers and other plant organs that promote exsudates with PA's precursors (Lamas &

Pérez, 1981), and they incorporate the alkaloid pharmacophagously whereas females obtain them

mostly by sperm transfer together with the nutrients used for egg production (Boggs, 1979). Some

species of Asteraceae and Boraginaceae may contain PA's and are known to be attractive for adult

Ithomiinae. These include Heliotropium sp. (Masters, 1968, Lamas & Pérez, 1981), Eupatorium sp.

(Trigo 1990; Brown, 1984a,b, 1987a; Pliske, 1975a,b; Edgar & Culvenor, 1976), and Epipendrum

orchids, for which Ithomiinae are the principal pollinators (DeVries and Stiles, 1990). PA's are used not

only for defences but also as precursors of pheromones, that are released by the males' hair patches.

The pheromones are used to attract females, but also function in territory recognition and defence,

leading other males to avoid invading an already occupied territory (Pliske, 1975b). Females transfer

PA's to their eggs, but larvae and pupae never contain the chemicals found in adults. Larvae are often

cryptic, greenish or brownish with yellow bars, whereas pupae are metallic or with golden shiny patches

(Brown & Freitas, 1994).

8


THE CASE OF OLERIA ONEGA SUBSPECIES

In this study interest was concentrated on two Ithomiinae subspecies: Oleria onega agarista (C.

Felder and R. Felder) and Oleria onega ssp. The latter is a recently discovered subspecies that has not

yet been described formally (Lamas, pers. com.). This new subspecies differs morphologically from O.

o. agarista by the presence of two white bands on the forewing that are never joined (in Oleria onega

agarista a transversal band connects this two bands on their middle part on the Cu1 and Cu2 veins)

(Lamas, pers com).

Oleria onega ssp. Oleria onega agarista

Both subspecies are endemic to the area of Tarapoto in north-eastern Peru, and are

geographically separated by a mountain chain, the Cerro Escalera: Oleria onega ssp. lives on the SW

side, whereas the NE side is the habitat of O. o. agarista. Two sites on the SW side were recorded as

"hybrid zones" where both O. o. ssp. and O. o. agarista occur in sympatry, with cases of hybridisation.

These morphological hybrids are recognised by an incomplete or absent transversal band (Lamas,

pers. com.). In the lowlands of the NE side a closely related Oleria species, Oleria lerida lerida, is

known and suspected to hybridize with O. o. agarista, but was too rare to be included in this study. Thus

our study will be concentrated only on the Cerro Escalera fauna, including the two Oleria onega

subspecies and their potential host plants.

The host plant used by the genus Oleria are Solanaceae (Drummond & Brown, 1987; Knapp

1997). Four Solanaceae species are suspected to be used by Oleria onega as host plants in the region

of Tarapoto ( Mallet, pers. comm.) : Solanum mite (Ruiz & Pav.), S. anceps (Ruiz & Pav.), S.

angustialatum (Bitter) and S. uleanum (Bitter). Their distributions differ on the two sides of the Cerro

9


Escalera: on the NE slope S. mite, S. anceps, and S. uleanum occur in sympatry, whereas on the SW

slope only S. mite is found. S. angustialatum grows on the upper part of the mountain. Oleria females

usually lay eggs singly on the Solanum leaves. However, field observations suggested that O o. ssp.

females might not oviposit on the host plant but on other objects such as stones, stems, dead leaves

and other plants adjacent to the host plant. The distance between the host plant and the oviposition

substrate may be up to one meter. When the female finds a host plant, two different reactions are

possible: the first is to quickly lay an egg under the leaf (mostly young leaves but not always), and the

second is to keep searching for a place on another substrate near the host plant. The behaviour of not

laying eggs on the host plant is mostly related to abiotic environmental factors in temperate climates

(Wiklund, 1984) and has not been previously reported in Ithomiinae. We suspected that this behaviour

may be related to differences in abundance of natural enemies. Therefore the oviposition behaviour of

both Oleria subspecies was studied and compared.

In the field of butterfly-host plant relationships, the most frequently studied subjects are host

specificity and host plant switches (Wiklund, 1975; Williams, 1983; Futuyma & Moreno, 1988;

Thompson, 1996), while specific oviposition behaviour in relation to external factors (climate predators

etc..) is less well studied. The few studies performed on this topic are mostly related to climate and host

plant abundance (Wiklund, 1984; Higashiura, 1989; Bergman, 1996; Steiner & Trusch, 2000). The

relation between adult oviposition preference, and on one hand larval host choice, as well as larval

success on the other, are closely related and both are taken into account in studies of plant-insect

relationship (Thompson, 1988; Nylin, 1993; Pires et al., 2000; Craig et al., 2000, Harris et al., 2000).

Here we will concentrate our interest on larval performance and behaviour, and the relation of these

parameters to female oviposition behaviour.

10


THESIS OUTLINE

The major aim of this study was to investigate the particular oviposition behaviour of Oleria

females. As we were not sure that the four Solanum species known from the literature are used as host

plant by the two butterfly subspecies, we tested the host plant preferences of both females and larvae,

as well as larval performance on the different Solanum species. Development of immature stage will be

described, as well as eggs, larval behaviour and larval survival, which are closely related to female

oviposition choice. However, it was also necessary to clarify the genetic relationship between the two

subspecies we studied and their morphological hybrids.

The four chapters treated in this thesis are closely linked. The first step is the study of

comparison of oviposition behaviour and host plant preferences for four Solanum species of the two

subspecies. These results led to further questions concerning larval behaviour, preferences,

performances and their relation with other environmental factors such as host plant density, diversity,

and predator fauna. Thereafter, the genetic relationships between the two subspecies were studied to

assert that we have two genetically different groups. As gene flow exist between the different

populations studied, we tried to understand the relationship between butterfly distribution and

abundance and their environmental factors, and to indicate sites where both subspecies may meet. As

a function of temperature and climate. GIS tools were used for cartographic representation of the

butterflies and host plant distribution, and Geospatial Analysis tools for the modelling of the different

environmental factors and butterfly density, allowing the study of their relations.

The present thesis focuses on the following questions:

1. Are the four Solanum species host plants of Oleria onega agarista and O. o. ssp? How does

oviposition behaviour of the two Oleria subspecies differ? Solanum mite, Solanum anceps, Solanum

angustialatum and Solanum uleanum are found in the area of the Cerro Escalera and are supposed

to be host plants, as the genus Solanum is usually associated with the genus Oleria (Drummond &

Brown, 1987; Knapp, 1997). In this study the aim was to evaluate the preferences of the females for

the four host-plants and thereafter to study the preferences between oviposition sites (on leaves of

the host plant or on other substrates) for both butterfly subspecies. We also wanted to understand the

11


factors inducing the behaviour of laying eggs close to the host plant through field observations and

laboratory experiments.

2. Is oviposition preference correlated with preference and performance of the larvae? Is there a

difference in potential predation pressure on larvae of both subspecies, that may affect their

behaviour? Is there differences in potential predators species for different oviposition environments?

Preference and performance of the larvae were tested on the four host plant species and correlated

with the results presented in Chapter 1. Survival of eggs in the field that were laid on the host plant

leaves was compared with survival of eggs laid on other substrates for both sides of the mountain.

This was done to investigate the potential relationship between egg survival and the laying of eggs

close to rather than on the host plant leaves. Larval behaviour and circadian activities were also

compared between both subspecies. The potential predator groups were identified and compared

between both sides of the mountain.

3. Are the two Oleria subspecies genetically distinct? The first aim was to define if Oleria onega ssp.

and Oleria onega agarista are genetically distinct, but also if differences occur between the different

populations studied, and if gene flow occurs. “Morphological hybrids” between O. o. ssp and O. o.

agarista were observed in the field, but we could not be sure that they were truly intermediate forms,

rather than a new differentiation of one subspecies. The different populations are geographically

separated by mountains or by deforested areas, but we did not know if the geographically close

populations were also genetically close. We selected RAPD techniques for a preliminary molecular

approach because they provide a virtually unlimited number of anonymous DNA markers (Williams et

al, 1990) and are therefore appropriate for initial, overall analysis of variation between populations.

4. Is there a relationships between distribution of Oleria and environmental factors such as altitude,

temperature, humidity and other factors? This aspect was investigated by GIS (Geographical

Information System) and Geospatial Analysis tools, allowing us to model the relationship between

butterfly abundance, temperature, humidity and other environmental factors. Even though the Cerro

Escalera range constitutes a biological barrier, individuals of O. o. agarista were found on the SW

side. Therefore we suppose that butterflies of the NE side had found a pathway around the mountain,

that depends on the topography, environmental and climatical factors. Through cartography of the

different factors, we wanted to find a relationship between organisms and their environment that may

induce the contact between subspecies. This part is divided in four main steps: (i) to build maps on

12


presence and absence of butterflies, the presence of the different host plant species, the soil type, the

tree size and the density of the understorey; (ii) thereafter, the mean temperature, humidity and

density of butterflies was measured for each site and was calculated by kriging to extend the data to

the whole study area (iii) conditional simulations allowed us to build probabilities maps (iv) results of

kriging, probabilities, landuse, topography and butterfly data were compared.

REFERENCES

Bates, H.W. 1862. Contribution of the insect fauna of Amazon Valley. Transactions of the Linnean

Society 23: 495.

Bergman, K.O. 1996. Oviposition, host plant choice and survival of a grass feeding butterfly, the

Woodland Brown (Lopinga achine) (Nymphalidae: Satyrinae). Journal of the research on

Lepidoptera???: 9-21.

Boggs, C.L.; Gilbert, L.E. 1979. Male contribution to egg production: first evidence for transfer of

nutrients at mating in butterflies. Science 206: 83-84.

Boppré, M. 1978. Chemical communication, plant relationships, and mimicry in the evolution of danaid

butterflies. Ent. exp. &appl., 24: 264-277.

Brown, K.S. 1979. Ecologia Geográfica e Evolução nas Florestas

Neotropicais (2 vol.). PhD dissertation. Campinas, Brazil.: Universidade Estadual de Campinas.

Brown, K.S. 1984a. Adult-obtained pyrrolizidine alkaloids defend ithomiine butterflies against a spider

predator. Nature 309 (5970): 707-709.

Brown, K.S. 1984b. Chemical ecology of dehydropyrrolizidine alkaloids in adult Ithomiinae (

Lepidoptera : Nymphalidae ). Rev. Brasil. Bio. 44, no 4: 435 - 460.

Brown, K.S. 1987a. Chemistry at the Solanaceae/Ithomiinae interface. Ann. Missouri Botanical Garden

74: 341-350.

Brown, K.S. 1987b. Biogeography and evolution of neotropical butterflies. In: T.C. Withmore and G.T.

Prance , ed. Biogeography and quaternary history in Tropical America. Oxford University Press. 66-104.

Brown, K.S.; Benson, W.W. 1974. Adaptative polymorphism Associated with multiple Müllerian

mimicry in Heliconius numata (Lepdi. Nymph.). Biotropica 6 ( 4): 205-228.

13


Brown, K.S., Freitas, A.V. 1994. Juvenile stages of Ithomiinae: overview and systematics

(Lepidoptera: Nymphalidae). Tropical Lepidoptera 5 (1).

Brown, K.S.; Trigo, J.R.; Rfancini, R.B.; Barros de Morais, A.B.; Motta, P.C. 1991. Aposematic

Insects on toxic host plants : Coevolution, colonization, and chemical emancipation. In: Peter W. Price

T, M. Lewinsohn, G. Wilson Fernandes, and Woodruff W. Benson, ed. Plant - animal interactions :

Evolutionary Ecology in Tropical and Temperate Regions.: John Wiley & sons, inc.

Brown, K.S.; Vasconcellos-Neto, J. 1976. Predation on aposematic Ithomiine Butterflies by Tanagers

Pipraeidea melanonota. Biotropica 8, no 2: 136 - 141.

Craig ,T.H.; Itami, J.K.; Shantz, C.; Abrahamson, W.G., Horner, J.D.; Craig J.V. 2000. The influence

of host plant variation and intraspecific competition on oviposition preference and offspring performance

in the host races of Eurosta solidaginis. Ecological Entomology 2000: 7-18.

DeVries, P.J.; Stiles, F.G. 1990. Attraction of Pyrrolizidine alkaloid seeking Lepidoptera to Epidendrum

paniculatum Orchids. Biotropica 22, no3: 290 - 297.

Drummond, B.A.; Brown, K.S. 1987. Ithomiinae (Lepidoptera, Nymphalidae): summary of known

larval food plants. Annals of the Missouri Botanical Garden 74: 341-358.

Edgar, J.A. 1982. Pyrrolizidines alkaloids sequestered by Solomon Island Danaine butterflies. The

feeding preferences of the Danainae and Ithomiinae. J. Zool. (London) 196: 385-399.

Edgar, J.A.; Culvenor, C.C.J. 1974. Pyrrolizidine ester alkaloid in danaid butterflies. Nature 248: 614-

616.

Futuyma, D.J.; Moreno, G. 1988. The evolution of ecological specialization. Annual Review of Ecology

and Systematics 19: 207-233.

Harris MOS, M.; Griffin W. 2001. Oviposition preferences of the Hessian fly and their consequences

for the survival and reproductive potential of offspring. Ecological Entomology 26: 473-486.

Higashiura, Y. 1989. Survival of eggs in the gypsy moth Lymantria dispar II. Oviposition site selection

in changing environments. Journal of Animal Ecology 58: 413-426.

Knapp, S., Helgason, T. 1997. A revision of Solanum section Pteroidea: Solanacea. Bull. nat. Hist.

Mus. London 27, no1: 31 - 73.

Lamas, G.; Pérez, J.E. 1981. Danainae e Ithomiinae (Lepidoptera: Nymphalidae) atraidos por

Heliotropium (Boraginaceae) en Madre de Dios, Peru. Revista Peruana Entomologica 24: 59-62.

Mallet, J.L.B. 1986. Gregarious roosting and home range in heliconius butterflies. Natl. Geogr. Res. 2:

14


198-215.

Masters, J.H. 1968. Collecting Ithomidae with Heliotrope. Journal of the Lepidopterists' society 22, no

2: 108 - 110.

Müller, F. 1878. Ueber die Vortheile der Mimicry bei Schmetterlinge. Zool. Anz. 1: 54-55.

Müller, F. 1879. Ituna and Thyridia: a remarkable case of mimicry in Butterflies. Kosmos: 100-109.

Nylin, S.; Janz, N. 1993. Oviposition preference and larval performance in Polygonia c-album

(Lepidoptera: Nymphalidae): the choice between bad and worse. Ecological Entomology 18: 394-398.

Pires, C.S.S.; Price, P.W.; Fontes, G. 2000. Preference-performance linkage in the neotropical

spittelbug Deois flavopicta, and its relation to the Phylogenetic Constraints Hypothesis. Ecological

Entomology 25: 71-80.

Pliske, T.E. 1975a. Courtship behavior and use of chemical communication by males of certain species

of Ithomiine butterflies ( Lepidoptera : Nymphalidae ). Annals of the entomological society of America

68, no 6: 935 - 942.

Pliske, T.E. 1975b. Pollination of Pyrrolizidine alkaloid - containing Plants by male Lepidoptera.

Environmental Entomology 4, no 3: 474 - 479.

Steiner, V.R.; Trusch, R. 2000. Egg-laying behaviour of Hipparchia statilinus in Eastern Germany

(Lepidoptera: Nymphalidae : Satyrinae). Stuttgarter Beitraege zur Naturkunde Serie a 606: 1-10.

Thompson, J.N. 1996. Trade-offs in larval performance on normal and novel hosts. Entomologia

Experimentalis et Applicata 80: 133-139.

Trigo, J.R.; Brown, K.S. 1990. Variation of pyrrolizidine alkaloids in Ithomiinae: a comparative study

between species feeding on Apocynaceae and Solanaceae. Chemoecology 1: 22-29.

Vasconcellos-Neto, J. 1980. Dínamica de populaçoes de Ithominae (Lepidoptera, Nymphalidae) em

Sumaré, SP. Campinas: Universidad Estadual de Campinas. 206.

Wiklund, C. 1975. The evolutionary relationship between adult Oviposition preferences and larval host

plant range in Papilio machaon L. Oecologia 18: 185-197.

Wiklund, C. 1984. Egg.laying patterns in butterflies in relation to their phenology and the visual

apparency and abundance of their host-plants. Oecologia 63: 23-29.

Williams, K.S. 1983. The Coevolution of Euphydryas chalcedona Butterflies and their larval host plants.

Oecologia 56: 336-340.

15

Chapter 1

Host-plant preference and oviposition behaviour of

two Oleria onega subspecies (Ithomiinae,

Lepidoptera) in north-eastern Peru.

Based on:

Gallusser, S.; Rahier, M.

Host-plant preference and oviposition behaviour of two Oleria onega subspecies

(Ithomiinae, Lepidoptera) in north-eastern Peru. In preparation for submission to Biological

Journal of Linnean Society

Chapter 1 Oviposition behaviour of two Ithomiinae subspecies

INTRODUCTION

Butterflies are among the most studied groups of insect in relation to their oviposition behaviour

and host-plant selection. Gilbert and Singer (1975) reviewed the principal traits of their ecology, Chew

and Robbins (1984) their egg laying behaviour, Thompson and Pellmyr (1991) the evolution of their

oviposition behaviour, Jaenike (1990) used them as model organism to study host specialisation and

genetic and non-genetic causes of variation in host specialisation. Despite of the large number of

publications on these subjects, new behavioural traits are still being found (Karlsson, 1995).

Several factors can act to shape selection on oviposition behaviour. Among these, larval

performance and survival, availability of adult resources (nectar or mating sites), distribution of

predators with respect to potential host plants, plant morphological characteristics, and density and

dispersion of host plants have all been discussed in the literature (Rausher, 1978, 1979; Tiritilli &

Thompson, 1988; Nylin & Janz, 1993) revealing that larval success is not always the main pressure for

female decision. To recognise a plant suitable for its offspring, the female uses a variety of cues. The

plant recognition process followed by the gravid female is similar in most butterfly species, and can be

divided in two main components (Hern et al., 1996; Renwick & Chew, 1994):

a) Pre-alighting behaviour : consists of searching, orientation and encounter: Females are

attracted to their host plant by chemical cues (Stadler et. al, 1995; Oyeyele, 1990; Pereyra & Bowers,

1988; Honda, 1990) and / or visual cues such as leaf shape and colour (Gilbert, 1975; Rausher, 1978;

Forsberg, 1987). Generally the balance between positive and negative chemical signals is important,

and therefore acceptance is not typically dictated by only one substance. The colour of the plant also

reveals its physiological condition. Butterflies which use visually apparent host plants usually seem to

find them without alighting on non-hosts (Wiklund, 1984).

b) Post-alighting discrimination characterised by landing, plant evaluation and acceptance or

rejection. Landing on a plant by a gravid female is the transition between pre-and post-alighting

behaviour. The contact evaluation on the plants' surfaces proceeds very rapidly, and involves sensory

receptors located on of the forelegs, proboscis, antennae and ovipositor (Renwick and Chew, 1994). In

some butterflies such as Nymphalids, the forelegs are reduced to small appendages used only as tactile

organs and chemosensory (Calvert, 1974).

16


Once a host plant is accepted, females have to find the best place on (or near) the plant where

they can lay their eggs; nevertheless they are often found to oviposit on sites not optimal for the fitness

of their offspring (Rausher, 1979; Chew & Robbins, 1984; Higashiura, 1989). In fact some butterflies are

known not to oviposit on their host plants: for example, some grass-feedings satyrids drop their eggs on

the ground with or without previous plant recognition ( Bergman, 1996; Steiner & Trusch, 2000)

whereas females of Parnassius apollo lay eggs at 1 to 2 m away from the host plant, sometimes without

previous recognition of the host plant (Wiklund, 1984). This is a common pattern in temperate areas for

butterfly species that spend the winter in the egg stage and that have herbs as host plants. Up to now,

the behaviour of laying eggs away from the host plant has not been reported for tropical butterflies.

We examined in this study the oviposition behaviour of two Ithomiinea butterfly subspecies.

Some early studies on Ithomiinae concern their ecology (Haber, 1978; Drummond, 1976; Drummond &

Brown, 1987; Masters, 1968). Recently, interest was focused on their mimetic qualities (DeVries et al.,

1997, Devries & Lande, 1999; Beccaloni 1997), biology (Brown, 1994), and their chemical particularities

(Brown, 1976 1984a,b, Brown & Vasconcellos-Neto1987, Cardoso, 1997, DeVries et al., 1997, Pinheiro

1996), but oviposition behaviour of these butterflies has been neglected. Oviposition behaviour differs

strictly between Oleria onega agarista (Felber & Felber) and Oleria onega ssp. nov. (a recently

discovered subspecies). Females of Oleria onega ssp. have been observed to recognise the plants by

alighting on them, but they search in the neighbourhood for other places to deposit their eggs (Mallet,

pers. comm.), probably taking into account a variety of unknown parameters, whereas the scarce

observations done on O. o. agarista suggest that females oviposit directly on the leaves of the host

plant.

Oleria onega is a complex of some fifteen subspecies (Lamas, pers. comm.) found throughout

the South-American tropical forests. Some subspecies have not yet been described in detail. Two of

these subspecies, investigated in this study, are present in the area of Tarapoto (S 06°22'50'; W

076°26'23''), in northeastern Peru: O. o. agarista and O .o. ssp., a recently discovered subspecies

whose status is still not clearly defined (Lamas, pers. comm.). Two other subspecies are present near

the area, but are too rare to be taken into account in this study. In some localities (Ahuashiyacu and

Shapaja, see below for site information) (Fig. 1) they are sympatric, and may hybridise (Lamas, pers.

comm.) (Chapter 3). However, these two subspecies are mostly geographically separated by a

mountain chain, the “Cerro Ecalera”, which constitutes a strong biological barrier between areas with

17


considerably different faunas and floras (Joron 2000; Joron et. al. 2001, Shulte 1999, Mallet, 1989,

1993); O. o. ssp.. is mostly found on the south-western slope, whereas O. o. agarista extends its range

on the north-eastern slope (thereafter termed as SW and NE slopes) and all the Huallaga Valley. Four

species of Solanaceae are expected to be used by Oleria onega as host plants (Drummond & Brown,

1987; Knapp & Helgason, 1997) : Solanum mite (Ruiz & Pav.), S. anceps (Ruiz & Pav), S.

angustialatum (Bitter), and S. uleanum (Bitter) (Knapp & Helgason, 1997). On the NE slope, three of

them occur in sympatry: S. mite, S. anceps, and S. uleanum, whereas on the SW slope only S. mite is

found. S. angustialatum grows on the upper part of the mountain. In the O. o. ssp. habitat (SW slope),

where only Solanum mite is found, the abundance of host-plant (in the sense of both density and area

covered) is also higher than on the other side of the mountain. Plants of S. mite grow in secondary

vegetation, mostly on the borders of the paths, and as deforestation is greater on the SW slope, open

areas are wider and allow bigger plant patches.

The behaviour of laying eggs on places other than the host plant had not been previously

observed in Ithomiinae. The evolution causes and the costs and benefits for the females and their

offspring, are unknown.

In this study the first aim was to examine host-plant preferences and host location behaviour of

each subspecies. We also evaluate how density and diversity of the Solanum affect the oviposition

behaviour, oviposition choice and the proportion of eggs laid on or near the plant. This is relevant here

because habitat occupied by the two subspecies exhibit a marked difference in diversity of potential

host plants. We document that O. o. ssp. lay eggs mostly on substrate other than plants and that the

preferred host plant for both subspecies may be Solanum mite as being the commonest and most

abundant one. Our second aim was to describe and quantify the differences in oviposition behaviour

between these two butterfly subspecies. Hypotheses to explain oviposition away from the host plant will

be proposed.

18


MATERIALS AND METHODS

Study organisms: Oleria onega agarista (Felder and Felder, 1862) and Oleria onega ssp. are

considered as conspecifics (Lamas, pers. comm.). O.o. ssp.. differs morphologically from O. o. agarista

by a narrower black edge of the wings, and by the presence of two white bands on the forewing that are

never joined (Lamas, pers com). Whereas in Oleria onega agarista a transversal band connects these

two bands on their middle part on the Cu1 and Cu2 veins. Morphological intermediates were recognised

by an incomplete or absent transversal band. Eggs are laid singly after plant recognition and this stage

lasts on average three days. Five larval instars are followed by pupation. Larval development requires

approximately seventeen days and pupation seven to eight days. Oleria may reached some six to eight

generation per year.

Study sites (fig. 1): The study sites are situated between Tarapoto (SW slope) and Km 30 of

the road (Carretera marginal) to Yurimaguas (NE slope) (S 06°24- 06°28 and W 76°18 – 76°22). The

road crosses the Cerro Escalera mountain by a tunnel (alt. 1100 m), and there is a great contrast in

climate between the two sides of the mountain. The SW side is usually sunny and hot (mean temp. ca.

30°C), while in the other side, the climate is wet and rainy, including fog and cooler temperatures (mean

temp. ca. 27° C).

Figure 1 : Geographical distribution of the field sites of Oleria onega ssp., and Oleria onega agarista in

North-eastern Peru.

19


Females of Oleria onega ssp.. were collected and followed in three sites on the SW slope,

Shilcayo, Urahuasha, Ahuashiyacu and Oleria onega agarista was studied on two different sites on the

NE side, Km28 and Km30. In Ahuashiyacu, both subspecies meet, but very few O. o. agarista

individuals were observed. Because of their slow flight and the fact that their are found in low

vegetation, females are relatively easy to follow in the field, but even so vegetation sometimes hinders

long pursuits. In our field observation, each step of the recognition and oviposition behaviour will be

detailed : pre-alighting recognition, plant evaluation by tasting and egg laying.

Observation of females in the field

Preliminary observations on Ithomiinae (unpublished results) revealed that the best time for field

observations was in the morning and at the beginning of the afternoon, between 9.00h and 13.00h. A

total of ninety-one gravid females were followed during three field seasons (between Oct. 1998 and Jan.

2001), in five sites : Ahuashiyacu, Shilcayo, Urahuasha, Km 30 and Km28. The following information

was noted for each female: distance that the female flew between the recognised host plant and the

oviposition site, the minimum distance that the larva had to move to reach the host plant, the substrate

on which eggs were laid and the time laps taken by the female between host plant recognition and

oviposition. What we term the “distance of the larvae”, is the minimum distance that the larvae have to

wander between the support where it hatched and the nearest host plant. This distance was estimated

following the shortest way to reach the plant, supposing that the larvae walk on the ground, on sticks or

other objects between the support and their first food. The “distance of the female” is the shortest flying

distance between the host plant and the support where the female laid the egg. The distance of the

female and the distance of the larvae were compared for three O. o. ssp. populations and analysed by a

Kruskal Wallis test. The differences between the substrates chosen by both butterflies subspecies were

analysed with Fisher's exact test. The mean time taken by both Oleria subspecies between plant

recognition and oviposition were compared by a t-test (Sokal & Rolf, 1995).

Oviposition experiments with females in cages

Experiments in flight cages, conducted in Tarapoto, were realised to allow an easier

manipulation of the host plants and the butterflies, and to provide a more controlled environment to

20


define the host preferences and the oviposition sites preferences of the females. Butterflies were kept in

cages constructed with wood and wire-mesh (plastic and metallic) stapled to the wood. To reduce

mortality due to ants, the cages were standing raised above the ground. Cage size was 1m3, which

allowed the butterflies to fly freely. Constant access to small recipients with 10% sucrose solution was

insured in the cage. Heliotropium flowers were regularly put into the cage as a source of alkaloids

(PA’s), as well as bird dropping as an amino acid source for egg production.

The O. o. ssp. females used in cage experiments originated from the Urahuasha, Shilcayo and

Ahuashiyacu populations (on the SW slope), and the O. o. agarista females from the Km30 population

(on the NE slope).

Screening experiments

The screening experiments were conducted in order to study the place on or away from the host

plant where eggs were laid, and to define clearly whether the two Oleria subspecies have different egg-

laying behaviours.

Five to fifteen individuals of each host plant species were collected and potted to facilitate their

manipulation. Females of O. o. ssp. (from Shilcayo) and O. o. agarista (from Km30) were separated in

two cages, with 10-15 females per cage. Within each cage, one individual of each of the four plant

species (Solanum mite, S. anceps, S. angustialatum and S. uleanum) was placed in each corner. Every

two days, eggs were collected and counted, and we recorded the number of eggs laid on each host

plant, or on objects in their vicinity (i.e., within a distance of 40 cm of the plant) (Schöps & Hanski,

2001). The radius circle of 40 cm, was chosen as at least 80 % of the eggs were observed to be laid

within this distance of the plant in the field.

Female preference:

The relative preferences of each butterfly subspecies for the two most common host plants, S.

mite and S. anceps, was tested with a single female per cage. In each cage, we placed two host plants

(one of each species) and two non-host plants, randomly selected in the habitat used as controls. The

plants were placed in each corner of the cage. Their positions were randomly distributed but an

alternation between host and non-host plant was maintained. The plants used as controls aid in

determining if a female recognises a host plant and oviposits on others or not. Every day, plants were

21


changed in the cage, so that each day females were offered a new treatment. Females were also

changed daily. However because not enough females were available, some of them were used two or

three times. During 14 days, eggs were collected each morning, and the position of each egg on or near

the plant was registered. Eggs encountered on other substrates within 40 cm from a plant, were

considered as associated with this plant. Data were analysed for each butterfly subspecies, applying

Kruskal Wallis test (Sokal & Rohlf, 1995) to test for differences in proportions of eggs layed on S. mite

and S. anceps. Differences between the number of eggs laid on leaves and on other objects was

analysed with Mann-Whitney test (Sokal & Rohlf, 1995) for each plant.

Effect of host plant density and diversity on female oviposition choice

As host plant diversity and density differed between the NE and SW sides of the Cerro

Escalera, two series of experiments were conducted in order to determine whether host plant density

and species diversity affect on female oviposition choice and the number of eggs laid per female. Host-

plant biomass density was measured for habitat on the two sides of the mountain (unpublished data).

On the South-western side, only Solanum mite was found, growing in big patches (of 20 to 100 plants),

and showing a higher leaf density per square meter than in the patches on the NE slope, where a

higher species diversity was found with Solanum mite, S. anceps and S. uleanum. The SW slope was

therefore characterised by a high density and low diversity, whereas the NE slope was characterised by

a low density and high diversity. Patches were also smaller (1 to 10 plants) on the NE slope. For our

experiments we used several treatments with different numbers of plants and different proportions of

the various species. For all the experiments described here females from the Shilcayo (SW) and Km30

(NE) populations were used, being changed every two days. However, because few females were

available for the experiment, some of them were re-tested using different treatment. Plants were also

changed each two days. Eggs were collected, and we recorded whether they were laid on the host plant

(and on which plant species) or on objects near ( up to 40 cm) the host plant. Eggs encountered on

other substrates within 40 cm from a plant, were considered as belonging to this plant, allowing a

neutral zone of at least 20 cm (cage sides were 1m long) between the different plants (i.e. plant plus a

40 cm radius circle). No control (non-host plant) was used in these two experiments. Females which did

not lay eggs during the experiment were discarded.

22


Four treatments were designed using two densities of two plant species (S. mite and S.

anceps), with one treatment per cage. The treatments had the following plant composition: 4 Solanum

mite, 1 S. mite, 4 S. anceps, 1 S. anceps. For the treatments with four plants, one plant was placed in

each corner of the cage, and for the treatments with a single plant, the plant was placed on the centre of

the cage. Thus, four cages were used for each Oleria subspecies. The experiment was replicated ten

times with different females. Eighteen O. o. ssp. and nineteen O. o. agarista females were used. The

proportion of eggs on the leaves divided by the total number of eggs laid in each cage were calculated

for each treatment and analysed by a Kruskal Wallis and by a Mann-Whitney test.

Five "diversity" treatments were designed with the following proportions of plants: A = 4

Solanum mite, B = 3 S. mite + 1 S. anceps, C = 2 S. mite + 2 S. anceps, D = 1 S. mite + 3 S. anceps, E

= 4 S. anceps. For all these treatments one plant was placed in each corner of the cage. Five cages

were used for each Oleria subspecies. This experiment was replicated thirteen times with one treatment

in one cage. Fifteen O. o. ssp. and nineteen O. o. agarista females were used. For each of the

treatments and subspecies, the following quantities and proportions were calculated :

a : The total number of eggs laid on each treatment by each female. Differences between

number eggs laid on the five treatments were analysed by a Kruskal Wallis test for both butterflies

subspecies. For each treatment, the differences between the two butterflies subspecies were analysed

by Mann-Whitney tests.

b : Proportion of eggs laid on leaves, regardless of the plants species, out of the total number of

eggs laid for each treatment. Differences between the five treatments were analysed by a Kruskal

Wallis test for each butterflies' subspecies. Differences between the subspecies were analysed by a

Mann-Whitney test for each treatment.

c : Proportion of eggs laid on leaves of S. mite and S. anceps, out of the total number of eggs

laid for each treatment. For each of the treatments B, C and D, that included both plant species, a

Mann-Whitney test was performed to examine the difference between plants.

23


RESULTS

Observations of females in the field

Females ready to lay eggs are easily recognisable because of their slow flight and their plant

searching behaviour. They appear to recognise the host plant species first visually, then land and

examine the leaves quickly with their forelegs. However, females do not always accept the host plant by

the visual or chemical signals. Sometimes they touch all the leaves (of a host and a non-host plant) that

they find on their way and continue to fly until they find a suitable plant, suggesting the importance of

contact recognition. An estimation of 80 % of plant individual accepted was true for all the studied

Solanum species. Once they locate a suitable host plant individual (in a patch of ten plants, sometimes

only one is chosen by several females), they show one of two different responses: either they quickly

lay the egg on the under side of the leaf (mostly but not always young leaves), or they continue to look

for an ideal place near the plant. Sometimes they appear to be disoriented and have to go back to the

plant to taste it again before they find an acceptable substrate for their egg. The substrate can be dead

material near the ground, other plants (non-host), stems or rocks, but never higher than the host plant

(max 1.5 m). Eggs were never observed to be laid more than one meter away from the plant but even at

this distance, reaching their first meal might require a dangerous and long trip for the newly hatched 1

mm caterpillar .

The Oleria onega ssp. females, in Ahuashiyacu, Shilcayo and Urahuasha (79 obs.), laid their

eggs almost always near the host-plant, on stones, stems, dead leaves, sometimes even 1m away form

the host plant. In contrast, Oleria onega agarista females (12 obs) , in Km30 and Km 28, laid their eggs

on the undersides of the Solanum mite leaves. The mean time lapse between plant recognition and

oviposition was 2.7 minutes (stdv. = 0.17) for O. o. ssp. and 1.7 minutes for O. o. agarista (stdv. = 0.25),

but these two means were not significantly different (T = 1.78, p = 0.08). Figure 2 compares two

distances between oviposition site and host plant, for three different populations of O. o. ssp.. The

distances that the females flew showed more variation than the distance that the larvae had to wander.

There were no significant differences among the three populations (Kruskal-Wallis : p > 0.1 for the

distance of the larvae, and p > 0.3 for the distance of the female). The standard deviations were very

large, revealing great variation among individuals, and these probably led to the insignificant differences

among populations even if distances of Ahuashiyacu populations appear longer than those of the others

24


two. The preferred oviposition site for O. o. agarista was on the host plant (Fisher's exact test: p =

0.001) (fig. 3), nevertheless some eggs were laid on dead material. For O. o. ssp. the preferred sites

were dead material on the ground and some eggs were also laid on non-host plants (species randomly

chosen by the females). For both butterfly subspecies, laying eggs on other substrates always followed

a previous alighting on the nearest host plant.

0

5

10

15

20

25

30

35

Ahuashiyacu Shilcayo Urahuasha

O. o. ssp. populations

Dis

tanc

e (c

m)

Distance of the larvaeDistance of the female

Figure 2 : results from the observations on oviposition in the field for O. o. ssp. (n =79 obs.).

Bars indicate the mean distance travelled by the female between the recognised host plant and the support where

the egg is laid, and the estimated mean distance for the larvae to wander for three O. o. ssp. populations :

Ahuashiyacu, Shilcayo and Urahuasha. Differences between the mean distances of the larvae and the female were

not significant between populations (Kruskal - Wallis : p > 0.2 for both distances).

25


0102030405060708090

100

Deadmaterial

stone non-hostplant

S. mite

Pro

porti

on o

f egg

s (%

)

O. o. sspO. o. agarista

Fig 3 : Results from the oviposition observations in the field for O. o. ssp. (total number of eggs laid = 79) and O.

o. agarista (total number of eggs laid = 12 obs.).

Bars indicate the mean proportion of eggs laid by the females of each subspecies on the three main

supports (dead plant material, stone, non-host plant) and on the host plant, Solanum mite. Differences

between the preferences of Oleria subspecies for different substrates were highly significant (Fisher's

exact test : p = 0.001).

Oviposition experiments with females in cages

Screening experiments

Solanum mite was the most preferred host plant for both subspecies (fig 4 A and B). O. o.

agarista oviposited mostly on leaves or at least an approximately equal proportion of eggs on leaves

and other substrates, while Oleria onega ssp. laid more eggs on other substrates near the host plant.

The same pattern was fared for all plants except for S. uleanum which was used only by O. onega ssp.,

and eggs were laid predominantly on the leaves of the plant (fig. 4A). Because of the variable number of

females per cage, these results are not quantitative and no statistical tests were performed on them.

26


0

10

20

30

40

50

60

70

S. mite S. anc. S. ang. S.ul.

Host plant species

Tota

l num

ber o

f egg

sLeavesOther substrate

0

5

10

15

20

25

30

S. mite S. anc. S. ang. S.ul.Host plant species

Tota

l num

ber o

f egg

s

LeavesOther substrate

(B)(A)

Fig. 4 A et B: Results from a choice test conducted with several females (10 - 15) per cage with four host plant

species (Solanum mite, S. anceps, S. angustialatum, S. uleanum ) for (A) Oleria onega ssp., and (B) Oleria onega

agarista. No statistical tests were performed on these data due to the variable number of females per cage. Eggs

were laid on host plants or on other objects such as cage walls and the plant's support.

Female preference:

Both Oleria subspecies laid significantly more eggs on S. mite than on S. anceps (Kruskal-

Wallis test for O. o. ssp. H = 8.49 d.f. = 2; P = 0.014, and for O. o. agarista H = 8.03 d.f. = 2; P = 0.018).

No eggs were laid on the control plant (Fig. 5A and B).

Results of Mann-Whitney tests performed on the proportions of eggs laid on the host plant and

on objects near the host-plant (within a distance of 40 cm) by each butterfly subspecies, revealed a

significantly greater proportion of eggs laid on S. mite leaves by O. o. agarista (U = 4.5, p = 0.03). O. o.

ssp. laid eggs in similar proportion on the leaves and on other substrates for both host plants (O. o. ssp.

: S. mite: U = 3.5, p= 0.19, and S. anceps U = 6, p = 0.56). For O. o. agarista as well, no difference was

found between the proportion of eggs laid on the S. anceps leaves and on other substrates nearby (U =

12, p = 0.33).

27


0

10

20

30

40

50

60

S. mite S.anceps Control

Host plant species

Pro

porti

on o

f egg

s (%

)

LeavesObjects

0

10

20

30

40

50

60

70

80

S. mite S.anceps Control

Host plant species

Pro

porti

on o

f egg

s (%

)

LeavesObjects

*

(B)(A)

Figure 5 : Results from a choice test with single females of (A) O. o. ssp. and (B) O. o. agarista per cage (n = 24

obs. for O. o. ssp. and n = 36 obs. for O. o. agarista).

Females were offered two host plant species and two non-host plants as control. Eggs were laid on leaves and

other objects (bags and cage walls). Differences between the proportions of eggs laid on S. mite leaves, objects

near S. mite, on S. anceps leaves, and objects near S. anceps were significant for both Oleria subspecies (Kruskal

Wallis : Fig.5A: H = 8.49 d.f. = 2; P= 0.014 p.; Fig 5B : H = 8.03 d.f. = 2; P = 0.018). Differences between the

different proportions of eggs laid on leaves and objects for each host plant were only significant for O. o. agarista

on S. mite ( * indicates the significant differences) (Mann - Whitney: Fig. 5A : S. mite : U = 3.5, p= 0.19, and S.

anceps : U = 6, p = 0.56, Fig 5B : S. mite :U = 4.5, * p = 0.03, and S. anceps : U = 12, p = 0.33)

Effect of host plant density and diversity

Non-host plants were not necessary as controls in this experiment, as no eggs were found on

them in the previous experiments.

a) "Density" : four treatments tested : A = 4 Solanum mite, B = 1 S. mite, C = 4 S. anceps, D = 1 S.

anceps.

For both Oleria subspecies no significant differences were found in the proportion of eggs laid

on the leaves in different density treatments, regardless of host plant species (Kruskal Wallis test p >

0.2 for Oleria onega ssp. and Oleria onega agarista). For both plant species and both Oleria subspecies

differences between the proportion of eggs laid on leaves under the two host plant density were not

significant (Mann-Whitney test : O. o. ssp : between treatment A and B: p = 0.67, between treatment C

and D : p = 0. 96; O. o. agarista : between treatment A and B: p = 0.42, between treatment C and D : p

28


= 0. 35). For some treatments, not enough data could be collected, because females laid an insufficient

number of eggs, and replicates were not in similar proportions. Because of the non-significant results,

no further analyses were performed on these data.

b) "Diversity" : Five treatments tested : A = 4 Solanum mite, B = 3 S. mite + 1 S. anceps, C = 2 S. mite

+ 2 S. anceps, D = 1 S. mite + 3 S. anceps, E = 4 S. anceps.

For O. o. ssp. the total number of eggs laid on each treatment differed significantly with a

maximum on treatment C and a minimum on treatments A and E (Kruskal-Wallis test : H = 9.97 d.f. = 4;

P = 0.041) No significant differences were found for O. o. agarista (Kruskal-Wallis test : H = 6.84 d.f. =

4; P = 0.14) (Fig 6). The mean number of eggs laid in each treatment varied between 2.5 and 4.5 for O.

o. ssp., and between 2,2 and 3.8 for O. o. agarista. No significant differences were found between the

total number of eggs laid by the two butterfly subspecies for any of the treatments ( Mann-Whitney test:

p > 0.1 for each treatment ).

0

1

2

3

4

5

6

A B C D E

Tota

l num

ber o

f egg

s

O. o. ssp (Kruskal Wallis: p= 0.041)

O. o. agarista (Kruskal Wallis: p= 0.14)

Figure 6 : Results from a choice test with O. o. ssp. ( 15 females, n = 65 obs.) and O. o. agarista (19 females, n =

65 obs) with different proportions of the two host plants in five different treatments.

Five diversity treatments were designed, with one treatment in one cage with the following proportions of plants:

A = 4 Solanum mite, B = 3 S. mite + 1 S. anceps, C = 2 S. mite + 2 S. anceps, D = 1 S. mite + 3 S. anceps, E = 4 S.

anceps. The total number of eggs laid on leaves and other substrates was compared between the five treatments for

both butterflies subspecies with significant results only for O. o. ssp. (Kruskal Wallis : O. o. ssp. : H = 9.97 d.f. =

4; P = 0.041 and O. o. agarista : H = 6.84 d.f. = 4; P = 0.14). Differences between both subspecies were analysed

for each pair of treatment with no significant results ( Mann-Whitney : P > 0.1 for the five treatments).

29


The different plant proportions did not seem to affect considerably the proportions of eggs laid

on the leaves of the host plant on the different treatment (Fig 7). For both butterfly subspecies 60 - 85 %

of the eggs were found on the leaves. The results of the Kruskal Wallis test performed between the

proportions of eggs laid on leaves in the five treatments were non-significant for both Oleria subspecies

( O. o. ssp.: H = 4.09 d.f. = 4; P = 0.39 and O. o. agarista : H = 1.6 d.f. 4; P = 0.8 ) (Fig 7). The results of

the Mann-Whitney tests performed between the proportion of eggs laid on leaves in each treatments by

both Oleria subspecies were only significant for treatment D (U = 33, p < 0.01) (Fig 7) where O. o. ssp.

laid fewer eggs on the host plant leaves than on other substrates.

0

10

20

30

40

50

60

70

80

90

100

A B C D ETreatment

Pro

porti

on o

f egg

s on

leav

es (%

)

O. o. ssp. O. o. agarista

*

Figure 7 : Results from a choice with O. o. ssp. ( 15 females, n = 65 obs.) and O. o. agarista (19 females, n = 65

obs) test with different proportions of the two host plants in five different treatments

Five diversity treatments were designed, with one treatment in one cage with the following plant proportions: A =

4 Solanum mite, B = 3 S. mite + 1 S. anceps, C = 2 S. mite + 2 S. anceps, D = 1 S. mite + 3 S. anceps, E = 4 S.

anceps.

The proportions of eggs on the leaves of both host plant were compared between each treatments for both

butterflies subspecies with no significant results (Kruskal-Wallis p > 0.4 for both Oleria subspecies). Differences

between both subspecies were analysed for each treatment with significant result only for treatment D ( Mann-

Whitney : treatment D: U = 33, * p < 0.01, others treatments p > 0.1).

The differences between the proportions of eggs laid on the two host plant species were

significant in treatments B and C for both butterfly subspecies ( Mann-Whitney: O. o. ssp. : Treatment B

30


: U = 31 p < 0.01; treatment C : U = 21 p < 0.01; O. o. agarista : Treatment B : U = 26.5 p < 0.01;

treatment C : U = 17.5 p < 0.01) where S. mite was the most preferred plant (Fig. 8), whereas in

treatment D the two subspecies laid eggs in similar proportions on leaves of both host plant species (O.

o. ssp. : treatment D : U = 73 p = 0.555; O. o. agarista : treatment D : U = 72 p = 0.095). For Solanum

anceps the proportion of eggs on leaves increased markedly with number of plants in the cage;

however, O. o. ssp. showed a greatest preference for S. mite. The proportion of eggs on S. mite leaves

was high for treatment A, but similar between treatments B and C, and fall drastically in treatment D for

both subspecies. Nevertheless, Solanum mite was the most preferred host plant for both Oleria

subspecies. O. o. agarista. appeared to accept more readily the host-plant switch, and showed a

greater preference for S. anceps than O. o. ssp.

0102030405060708090

100

A B C D ETreatments

Egg

s pr

opor

tion

on le

aves

(%) Mite

anceps

**

Kruskal Wallis Pvalue = 0.14

0

10

20

30

40

50

60

70

80

90

A B C D ETreatments

Egg

s pr

opor

tion

on le

aves

(%) Mite

anceps

**

Kruskal Wallis Pvalue = 0.041(B)(A)

Figure 8 : Results from a choice test with different proportion of the two host plants in five treatments

with (A) O. o. ssp. and (B) O. o. agarista The difference between the proportion of eggs laid on the two host plants were significant for treatments B, and C,

(* indicate the significant results) but not for treatment D ( Mann-Whitney : Fig 8A : treatment B: : U = 31 p <

0.01; treatment C : U = 21 p < 0.01; and treatment D p > 0.5; Fig 8B : Treatment B : U = 26.5 p < 0.01; treatment

C : U = 17.5 p < 0.01 and treatment D p > 0.09).

31


DISCUSSION

oviposition preferences

The results from the screening experiment on oviposition preference showed that the four

presumed host plants were chosen by both subspecies of butterflies and, among these, Solanum mite

was their most preferred host plant. In the experiment with a single female per cage, no eggs were laid

on or near the control plants. It can thus be assumed that all eggs on the wall of the cage were laid after

recognition of the nearest host plant (within a distance of 40 cm.). In the screening experiment, a large

number of eggs was laid on S. uleanum by O. o. ssp. females; this may be due to chemicals similarities

between this plant and S. mite. However, there might be another explanation: it has been demonstrated

that when a female is more motivated, it is more likely to accept a non-preferred host plant for

oviposition ( Fitt, 1986, Singer et al., 1992, Hopkins & Van Loon, 2001). In our case, only one S. mite

individual was present in the cage, with several females laying eggs on it. The high competition in this

case may induce a highly motivated female to oviposit on a less accepted host. This remains a

hypothesis, as the motivation (i.e. number of eggs in the abdomen) of the female was not defined.

The hierarchy in host preference is similar in both subspecies. They first choose Solanum mite,

then Solanum anceps, and finally S. angustialatum and uleanum with no great difference. Nevertheless,

it might be kept in mind that all the females used in these experiments came from the field, and were

therefore not “naive”. In most oviposition preference studies, newly emerged and mated females are

used (Craig et al. 2000; Bernays, 1999; Schöps & Hanski, 2001; Hopkins & Van Loon, 2001) but in our

case no individuals born in the laboratory had accepted to mate. The results may be biased mostly for

O. o. agarista that naturally have S. mite, S. anceps and S. uleanum in their environment, whereas O. o.

ssp, only knows S. mite. However, learning appears not to have an effect on the results, as preferences

of both Oleria subspecies were quite similar.

Differences in preference for leaves and other substrates.

The egg laying behaviour of the two Oleria onega subspecies indicated differences in the choice

of oviposition sites, i.e. whether eggs were laid on host-plant leaves or on objects close by. Oleria

onega agarista oviposited mostly on the leaves, whereas O. o ssp. searched for other substrates to lay

eggs. When the eggs were laid on the host plant, no specialisation was observed on a particular site, as

32


occurs with other lepidopteran species (Benson 1978; Thompson 1983; Janzen 1988; Rausher 1979,

Chew & Robbins, 1984; Higashiura 1989). The known reasons for a within-plant specialisation are

climatic factors in cold regions (Higashiura 1989), protection against predators (Benson 1978;

Thompson 1983; Janzen 1988; Rausher 1979, Chew & Robbins, 1984) and quality of food for the

larvae (Barker & Maczka, 1996, Craig, 2000; Steiner & Trusch, 2000), even though female choice is not

always correlated with influence of oviposition site on offspring fitness. In the case of Oleria, females

can lay eggs on old leaves as well as on young meristems, stems and any other part of the plant. We

may conclude that the site on the plant does not affect the larval performance, and are not chosen to

avoid predators nor to provide the best food for the larvae. In contrast, oviposition on objects near the

host-plant suggests a high selection pressure against eggs, reducing their survival when laid on any

part of the plant. Costs seem to be higher for O. o. ssp., on the SW slope, as this behaviour was

scarcely observed in O. o. agarista, coming from the NE side. We presume that the main factor of costs,

inducing this behaviour is the presence of natural enemies of eggs and larvae, occurring more

frequently on the SW slope. The behavioural difference between the two subspecies was greater when

several females were in the same cage than in experiments with a single female per cage. This result

suggests that the fact of not laying eggs on the host plant might be more accentuated when higher

competition between females for oviposition sites is greater, and this may influence the behaviour and

preferences of all females. It has been reported that, for some species that usually lay eggs in clusters,

the size of the clutches decreases when the number of eggs-laying females per patch increases (Parker

& Courtney, 1984; Parker & Begon, 1986). As density of female was higher on the SW side than on the

NE side, the difficulty encountered by females in finding the best place can be compensated by a higher

egg survivorship and the high host plant density. In our case, when a high concentration of females

occurs, the existing behaviour of avoiding to oviposit on the host plant, may be influenced by a new

cost, which is competition.

Effect of density and diversity of plants

In the experiment conducted to test the effect of density there were no effects on oviposition

pattern but this result may be biased since females generally laid few eggs in this experiment, and

these results will not be taken into account here.

33


In the experiment with different host plant species composition, we could expect that the choice

will reduce the number of eggs laid in the mixed treatments, as observed in other study, due to the fact

that complex stimuli reduce "decisiveness" or allow distraction (Bernays, 1999; Hopkins and Van Loon,

2001). In our case, when the two host plants were present, unexpectedly more eggs were laid than in

the treatment with a single plant species. When only S. anceps, the host of low acceptability, was

present the number of eggs laid was similar to that when only S. mite was present. This may be

explained by the fact that when female have a "long" life time (Oleria live up to two months), they have

the possibility to wait when conditions for oviposition are not optimal. In our case, females laid more

eggs when they have the opportunity to choose the plants (in the mixed treatments). If they had a

shorter life they may lay eggs in the same quantities in all the treatment. However this hypothesis is

hard to apply in our case as we ignore the real age of the females used.

Therefore, even though S. mite was the most preferred host, a switch to other host plant (such

as S. anceps) may occur easily. This was clearly demonstrated by the non-significant result of the

Mann-Whitney test for treatment D for both butterfly subspecies. When one S. anceps was present with

three S. mite, this single plant was neglected whereas when three S. anceps were present with one S.

mite (treatment D), S. mite was still the more attractive (in the case of O. o. ssp.), but S. anceps was

also accepted as a host plant When both host plants were present in equal proportion, S. mite was the

most preferred by both butterfly subspecies. O. o. agarista seemed to accept S. anceps as a host plant

readily than did O. o. ssp.. This is logical, as both butterfly and plant are found in the same environment

in the field. The proportion of eggs laid on leaves was nearly the same for O. o. ssp. and O. o. agarista

(Fig 8). Both subspecies apparently reacted similarly to the different composition of plants in the cages,

although O. o. ssp. did not show as great difference in the choice of the oviposition site ( on and away

from the plant), as it did in the previous experiment. However it might be recorded that for this

experiment fewer females of O. o. ssp. were tested than for the other subspecies. Another problem with

this experiment was that few females were available, and therefore some of them were tested twice with

the same plant proportions (but plant specimens were changed) and could then learn to recognise the

plant species. However some studies have revealed that learning is greater in butterflies when foraging

than when laying eggs (McNeely & Singer, 2001) and that sometimes butterflies do not learn after

previous encounters with different host plant ( Schöps & Hanski, 2001).

34


How do these results fit with the natural composition of plants and mortality due to predators?

We supposed that Oleria onega ssp., which does not encounter S. anceps and S. uleanum in its

natural environment, would neglect them more than did O. o. agarista. However preferences of both

subspecies were similar. In some cases, O. o. ssp. showed a higher acceptability for new hosts. For

example in the screening experiment, the high number of eggs laid on S. uleanum by O. o. ssp. females

and could be related to the fact that some of them were highly stimulated to oviposit and that density of

S. mite in the cage was too low to satisfy the female constraining it to choose a new host. Even if any

switch to a new host involves genetical and learning parameters that take time to be acquired by a

female (Singer, 1983, 1986; Jaenike 1990; Thompson 1988; Thompson & Pellmyr, 1991, Wiklund,

1975), it has been noted that in a specialist population, when the preferred host is scarce, the shift to a

host of lower acceptability may be carried out quickly by some individuals that are more generalists,

whereas the majority of individuals of the population retain egg loads in case a superior resource

becomes available later (Wiklund, 1981, Hokins & Van Loon, 2001). In our case, a factor that could

facilitate host switching is that in the four tested plants, the chemical compounds should be similar as

the plants are closely related.

For the oviposition site preferences, we suppose that the main factor of costs inducing the

behaviour of not laying eggs on the host plant is the presence of natural enemies of eggs and larvae,

and eventually ovipositing females, occurring mostly on the SW slope (see Chapter 2). The climatic

conditions around the host plant and the food quality factors do not seem to have such importance to

induce a new oviposition strategy, however host plant are regularly cut by the farmers and this may be

another factor inducing the female to avoid oviposition on the plant. As this behaviour was accentuated

when many females were in the same cage, we suppose that this strategy, first induced by predation,

may be used thereafter by the butterflies as a way to reduce competition. In contrast, a factor that may

have helped the butterflies to develop this behaviour on the SW side, is the higher host-plant density

observed there, making food easier to find for a newly hatched larva.

35


CONCLUSION

The most pertinent result found in this study was that both cage experiments and field

observation revealed that the two butterfly subspecies have a distinct behaviour. O. o. ssp. lays most of

its eggs on objects close to the plant and O. o. agarista on the plant itself. It is important to note,

however, that in both cases, exceptions were observed. Another interesting result revealed by the

experiments conducted in the cages, is that Solanum mite is the most preferred host plant, but that

oviposition switches to other Solanum species occur easily when S. mite is rare or absent. Laying eggs

away from the plant could be a new behaviour that O. o. ssp. is developing, but that not all females

have already completely evolved. On the other hand a few O. o. agarista females seem to adopt this

behaviour. A possible explanation may be that for O. o. ssp. the costs due to predation on the host plant

induced the butterflies to find a new strategy, searching for the more secure place for the eggs. The

cost of oviposition away from the plant may be compensated by less competition for the ovipositing

females to find a place on the plant, and by the higher host plant density found on the SW side of the

mountain. The composition of predators should be different between both mountain sides, and one (or

several) predator species are not found on the NE slope. A second hypothesis, completely different,

may be that in the SW slope, there are more farmers and more deforestation, since at least fifteen

years. The large patches of Solanum mite found on the path-hedges, are regularly cut to clean the path.

As small plants are still present after cutting, it is possible that the butterflies have developed the

behaviour of not laying eggs on plants, hence allowing greater chances for the larvae to find their food

plants. This adaptation is possible due to high number of butterflies generations per year (at least six

generations) and due to the fact that deforestation became greater from fifteen years ago, that may

"learn" to avoid oviposition on endangered sites.

New questions arise from this study that we will partly be tried to answer in Chapter 2: how

suitable are the four host plants for the larvae? Are there correlations between female preferences, and

larval preference and performance? Are soil fauna and presence of potential predators on plants

different between both mountain sides? Are there inducing heavy costs, enough to a new behaviour to

arise?

36


ACKNOWLEDGEMENTS

This work was supported by a Grant from the Commission for travel of the Swiss Society of

Natural Science (SANW), and the Matthey-Wüthrich funds for travels. Thanks are due to Dr J. Mallet

(University College, London) for giving the subject of this study, to Dr. G. Lamas (Universidad Mayor de

San Marcos, Lima) for determination of Ithomiinae, and to Dr K. Gotthard (University of Neuchâtel), Dr.

B. Benrey (University of Neuchâtel), and Dr. Ch. Wiklund (University of Stockholm), for helpful

comments on the manuscript. Thanks are due to Dr. M. Joron, G. Valencia, M. Abanto, and J. Reategui

for field collaboration.

REFERENCES Barker, A.M.; Maczka, J.M. 1996. The relationships between host selection and subsequent larval

performance in three free-living graminivorous sawflies. Ecological Entomology 21: 317-327.

Beccaloni, G.W. 1997. Vertical stratification of ithomiine butterfly (Nymphalidae: Ithomiinae) mimicry

complexes: the relationship between adult flight height and larval host-plant height. Biological Journal of

the Linnean Society 62: 313-341.

Benson, W.W. 1978. Resource partitioning in passion vine butterflies. Evolution 32 (3): 493-518.

Bergman, K.O. 1996. Oviposition, host plant choice and survival of a grass feeding butterfly, the

Woodland Brown (Lopinga achine) (Nymphalidae: Satyrinae). Journal of the research on Lepidoptera

35 9-21.

Bernays, E.A. 1999. When host choice is a problem for a generalist herbivore: experiments with the

whitefly, Bemisia tabaci. Ecological Entomology 24: 260-267.


predator. Nature 309 (5970): 707-709.



Brown, K.S. 1987. Chemistry at the Solanaceae/Ithomiinae interface. Ann. Missouri Botanical Garden

74: 341-350.

Brown, K.S.; Freitas, A.V. 1994. Juvenile stages of Ithomiinae: overview and systematics

37


(Lepidoptera: Nymphalidae). Tropical Lepidoptera 5 (1): 9-20.

Brown, K.S.; Vasconcellos N., J. 1976. Predation on aposematic Ithomiine Butterflies by Tanagers


Calvert, V.H. 1974. The external morphology of foretarsals receptors involved with host discrimination

by the nymphalid butterfly Chlorocine lacinia. Annals of the Entomological Society of America 67: 853-

856.

Cardoso, M.Z. 1997. Testing chemical defence based on pyrrolizidines alkaloids. Animal Behaviour 54

(4): 985 - 991.

Chew, F.S.;Robbins, R. K. 1984. Egg-laying in Butterflies. In: Van Wright R, Ackery P. R., ed. The

biology of Butterflies. New York: New York Academic. 65-79.

Craig, T.P.; Itami, J.K.; Shantz, C.; Abrahamson, W.G.; Horner, J.D.; Craig, J.V. 2000. The

influence of host plant variation and intraspecific competition on oviposition preference and offspring

performance in the host races of Eurosta solidaginis. Ecological Entomology 25: 7-18.

DeVries, P.J.; Murray, D.; Lande, R. 1997. Species diversity in vertical, horizontal, and temporal

dimensions of a fruit feeding butterfly community in an Ecuadorian rainforest. Biological journal of the

Linnean society 62: 343 - 364.

Devries, P.J.; Lande, R. 1999. Associations of co-mimetic Ithomiinae butterflies on small spatial and

temporal scales in a neotropical rainforest. Biological Journal of the Linnean Society 67: 73-85.

Drummond, B.A. 1976. Comparative ecology and mimetic relationship of Ithomiinae butterflies in

eastern Ecuador. Florida: University of Florida.



Fitt, G.P. 1986. The influence of a shortage of hosts on the specificity of ovipositional behaviour in

species of Dacus (Diptera, Tephritidae). Physiological Entomology 11: 133-143.

Forsberg, J. 1987. Size discrimination among conspecific hostplants in two pierid butterflies : Pieris

napi L. and Pontia daplidice L. Oecologia 72: 52-57.

Gilbert, L.E.; Singer, M. C. 1975. Butterfly ecology. Annual Review of Ecological Systematic 6: 365-

397.

Haber, W.A. 1978. Evolutionary Ecology of Tropical Mimetic butterflies: Function of Ithomiine scent

scales in aggregating behaviour: Uni. of Minnesota.

38


Hern, A.; Edwards-jones, G.; McKinlay, R.G. 1996. A review of the preoviposition behaviour of the

small cabbage white butterfly, Pieris rapae (Lepidoptera, Pieridae). Annals of appplied biology, Warwick

128 (2): 349-371.

Higashiura Y. 1989. Survival of eggs in the gypsy moth Lymantria dispar II. Oviposition site selection in

changing environments. Journal of animal ecology 58: 413-426.

Honda, K. 1990. Identification of host-plant chemicals stimulating oviposition by swallotail butterfly,

Papilio protenor. Journal of Chemical Ecology 16 (2): 325-337.

Hopkins, R.J.; VanLoon, J.J.A. 2001. The effect of host acceptability on oviposition and egg

accumulation by the small white butterly, Pieris rapae. Physiological Entomology 26: 149-157.

Jaenike, J. 1990. Host specialization in phytophagous insects. Annual Review of Ecological Systematic

21: 243 - 273.

Janzen, D.H. 1988. Ecological characterization of a Costa Rican dry forest caterpillar fauna. Biotropica

20: 120-135.

Joron, M. 2000. Coloration avertissante et mimetisme müllérien: le problème de la diversification

Thèse, Université des Sciences et Technique du Languedoc. Montpellier: Académie de Montpellier.

Joron, M.; Wynn, I.R.; Lamas, G.; Mallet, J.L.B. 2001. Variable selection and the coexistence of

multiple mimetic forms of the butterfly Heliconius numata. Evolutionary Ecology 13: 721-754.

Karlsson, B. 1995. Resource allocation and mating systems in butterflies. Evolution 49(5: 955-961.

Knapp, S.; Helgason, T. 1997. A revision of Solanum section Pteroidea: Solanacea. Bull. nat. Hist.

Mus. London 27, no1: 31 - 73.

Mallet, J.L.B. 1989 The genetics of warning color in peruvian hybrid zones of Heliconius erato and

Heliconius melpomene. Proc. R. Soc. Lond. B. Biol. Sci. 236: 163-185.

Mallet, J.L.B. 1993. Speciation raciation and color pattern evolution in heliconius butterflies evidence

from hybrid zones. In: Harrisson, R.G. ed. Hybrid zones and the Evolutionary process. New - York:

Oxford university press, inc.pp. 226-260.

Masters, J.H. 1968. Collecting Ithomidae with Heliotrope. Journal of the Lepidopterist's society 22, no

2: 108 - 110.

McNeely, C.; Singer, M.C. 2001. Contrasting the roles of learning in butterflies foraging for nectar and

oviposition sites. Animal Behaviour 61: 847-852.

Nylin, S.; Janz, N. 1993. Oviposition preference and larval performance in Polygonia c-album

39



Oyeyele, S.O.; Zalucki, M.P. 1990. Cardiac glycosides and oviposition by Danaus plexippus on

Asclepias fructicosa in south-east Queensland (Australia), with notes on the effect of plant nitrogen

content. Ecological Entomology 15: 177-185.

Parker, G.A.; Courtney, S.P. 1984. Model of clutch size in insect oviposition. Theorical Population

Biology 26: 27-48.

Parker, G.A.; Begon, M. 1986. Optimal egg size and clutch size, effect of environment and maternal

phenotype. American Naturalist 128: 573-595.

Pereyra, P.C.; Bowers, M.D. 1988. Iridoid glycosides as oviposition stimulants for the buckeye

butterfly, Junonia coenia (Nymphalidae). Journal of Chemical Ecology 14: 917-928.

Pinheiro, C.E.G. 1996. Palatability and escaping ability in Neotropical Butterflies: tests with wild

Kingbirds (Tyrannus melancholicus, Tyrannidae). Biological Journal of the Linnean Society 59: 351-365.

Rausher, M. 1978. Search image for leaf shape in butterflies. Science 200: 1071-1073.

Rausher, M. 1979. Larval habitat suitability and oviposition preference in three related butterflies.

Ecology 60 (3): 503-511.

Renwick, J.A.; Chew, F.S. 1994. Oviposition behavior in Lepidoptera. Annual Review Entomology 39:

377-400.

Schöps, K.; Hanski, I. 2001. Population level correlation between pre-Alighting and post-alighting host

plant preference in the Glanville fritillary butterfly. Ecological Entomology 26: 517-524.

Schulte, R. 1999. Die Pfeilgiftfrösche- Artenteil, Peru (Vol. 2). INIBICO- Waiblingen, Germany,

Waiblingen.

Singer, M.C. 1983. Determinants of multiple host use by a phytophagous insect population. Evolution

37: 389-403.

Singer, M.C. 1986. The definition and measurement of oviposition preference in plant feeding insects.

In: T.A. Miller JAM, ed. Insect-plant interactions. New-York: Springer-Verlag. 65-94.

Singer, M.; Vasco D.; Parmesan, C.; Thomas, C.D.; Ng, D.1992. Distinguishing between "preference"

and "motivation" in food choice, an example from insect oviposition. Animal Behaviour 44: 436-471.

Sokal, R.R.; Rolf, F.J. 1995. Biometry : the principles and practice of statistics in biological research,

3rd edition. W.H. Freeman and Co eds. New-York.

Stadler, E.; Renwick, J.A.A.; Radke, C.D.; Sachdev-Gupta, K. 1995. Tarsal contact chemoreceptor

40


response to glucosinolates and cardenolides mediating oviposition in Pieris rapae. Physiological

Entomology 20 (2): 175-187.

Steiner, V.R.; Trusch, R. 2000. Egg-laying behaviour of Hipparchia statilinus in Eastern Germany

(Lepidoptera: Nymphalidae : Satyrinae). Stuttgarter Beitraege zur Naturkunde Serie a 606: 1-10.

Thompson, J.N. 1983. Selection pressure on phytophagous insects feeding on small host plant. Oikos

40: 438 - 444.

Thompson, J.N. 1988. Evolutionary ecology of the relationship between oviposition preference and

performance of offspring in phytophagous insects. Entomology Experimentalis and Applicata 47: 3-14.

Thompson, J.N.; Pellmyr, O. 1991. Evolution of oviposition behavior and host preference in

Lepidoptera. Annual Review Entomology 36: 65-89.

Tiritilli, M.E. Thompson, N.J. 1988. Variation in swallowtail/plant interactions : host selection and the

shapes of survivorship curves. Oikos 53: 153 - 160.

Wiklund, C. 1975. The evolutionary relationship between adult oviposition preferences and larval host

plant range in Papilio machaon L. Oecologia 18: 185-197.

Wiklund, C. 1981. Generalist vs. specialist oviposition behaviour in Papilio machaon (Lepidoptera) and

functional aspects on the hierarchy of oviposition preferences. Oikos 36: 163-170.



41

Chapter 2

Comparison of larval behaviour of two Oleria onega

subspecies (Ithomiinae, Lepidoptera), in relation to

predators of their early stages.

Based on:

Gallusser, S.; Rahier, M.

Comparison of larval behaviour of two Oleria onega subspecies (Ithomiinae, Lepidoptera),

in relation to predators of their early stages. In preparation for submission to Ecological

Entomology.

Chapter 2 Larval behaviour of two Ithomiinae subspecies

INTRODUCTION

In theory, the successful development of the larvae should be the main factor dictating the

female oviposition choice however, in nature, other environmental factors are also known or suspected

to play an important role, such as: i) availability of resources for adults (food and mating sites), ii) factors

influencing temperature and/or light intensity (shady versus open sites), iii) distribution of predators with

respect to potential host plants, iv) plant morphological characteristics such as size or shape which may

be correlated with larval success, and v) density and dispersion of host plants. In some cases females

oviposit in places that are not advantageous for their offspring (Rausher 1979, Chew & Robbins, 1984,

Higashiura, 1989). Indeed, numerous studies have compared the preferences of ovipositing females for

certain host plants with the performance of the larvae, revealing that in most of the cases this

correlation is unexpectedly low (Williams, 1983; Pires et al., 2000; Craig et al., 2000, Harris et al.,

2000). In exchange, other cases have shown that correlation may be very high (Nylin & Janz, 1993;

Barker & Maczka, 1996). However, study on host preferences can lead to a confusion as discrimination

within host species can mask or confound discrimination among species (Singer & Lee, 2000). For both

female preference and larval performance, we can observe a selection among the plants chosen by the

females and those that the larvae are favoured, and ranking of host plant is not always the same for

females and their offspring (Harris et al., 2001). Many factors may interact to determine preferences and

performances of both gravid females and their offspring, and a failure to consider important factors, or

the interactions among factors, may lead to an incomplete understanding of the preference-

performance relationship. However, discrepancies may also be due to conflicting selection pressures

acting on ovipositing females, with for example, increased predation on higher quality hosts leading to a

lack of directional selection (Clancy, et al. 1993), or to avoidance of oviposition on the plant (Inouye &

Taylor, 1979; Freitas & Oliveira, 1996).

Some butterflies are known not to oviposit on their host plants. However this particularity has

been observed mostly in species that overwinter in the egg stage and whose host plants are

superabundant (Wiklund, 1984). In the tropics, the particularity of ovipositing on objects near the host

plant has not yet been observed. In contrast to the observed cases in temperate regions, predation is

likely to be one of the possible factors that may induce this behaviour as environmental factors such as

temperature and humidity do not considerably vary in time and space at a small scale in a tropical

42


forest. Previous studies on ants have shown that predation on larvae may be reduced by larval defence

mechanisms but also by behaviour of the gravid female (Inouye & Taylor, 1979; Freitas & Oliveira,

1996). Larval defences may involve scoli, spines and hairs (Frost, 1959), unpalatability (Brower, 1984)

camouflage and mimicry (Edmunds, 1974), myrmecophily (DeVries, 1991) and the construction of frass

chains (Machado & Freitas, 2001). In contrast, survival of the egg stage may not depend exclusively on

intrinsic egg protection factors (unpalatability, camouflage) but also on female oviposition behaviour.

The aim of this study is to investigate the larval biology of two Ithomiinae subspecies that are

endemic to north-eastern Peru: Oleria onega agarista (Felber & Felber) and Oleria onega ssp. a

recently discovered subspecies. Interest in Ithomiinae was first focused on their mimetic particularities

(Bates 1962, Müller 1976, Brown & Vasconcellos-Neto, 1976; Brown, 1979) and on their chemical

relations with plants (Brown, 1984 a, b,). Relations between Ithomiinae and their larval host plants were

described at the genus level by Brown (Brown, 1987) who showed that the genus Oleria is associated

with Solanum species. However, the larval biology of the different species remain patchily studied and

that only a few of species have been described in more detail by Young (Young, 1972, 1973, 1974a,b,c,

1977, 1978a,b) and by Haber (1978). More recently, juvenile stages of most species of the genus were

described and photographed by Brown (Brown & Freitas, 1994). Field observations in the north-eastern

Peru, near the town of Tarapoto (S 06°22'50'; W 076°26'23''), suggested that four Solanum species are

probably used by Oleria onega subspecies as host plants (Drummond & Brown, 1987; Knapp &

Helgason, 1997): S. mite (Ruiz & Pav.), S. anceps (Ruiz & Pav.), S. angustialatum (Bitter) and S.

uleanum (Bitter). The “Cerro Escalera” is a mountain chain which constitutes a strong biological barrier

between areas with considerably different fauna and flora (Joron 2000; Joron et al., 2001; Schulte 1999,

Mallet, 1989, 1993). O. o. ssp. is mostly found on the south-western slope, whereas O. o. agarista

occurs on the north-eastern slope (abbreviated below as SW and NE slopes) and in the Huallaga

Valley. Both subspecies occur at 200 - 800m altitude. On the NE slope, three of the four Solanaceae

species occur in sympatry: S. mite, S. anceps, and S. uleanum, whereas on the SW slope only S. mite

is found. S. angustialatum grows on the upper part of the mountain (≈1'100 m alt.). In the habitat of O.

o. ssp., where only Solanum mite is found, the abundance of this host-plant (i.e. density and area

covered) is higher than on the other side of the mountain. S. mite grows abundantly in secondary

vegetation after deforestation, which is greater on the SW slope. Therefore open areas are greater in

size and allow for bigger plant patches, mostly on the path edges (Knapp & Helgason, 1997).

43


Female oviposition preferences were tested in a previous study (Chapter 1) that revealed a

hierarchy among the four plants: S. mite, the most abundant and the most preferred host, followed by S.

anceps and subsequently by S. angustialatum and S. uleanum, which do not differ in their

attractiveness. Nevertheless the influence of different plants on larval fitness remains unknown.

Females of the genus Oleria generally lay their eggs singly on the underside of the leaves of their host

plant. However, Oleria onega ssp. females showed the particularity of laying most of their eggs on

objects near rather than on the host plant (Chapter 1). After successful recognition of the host plant,

females flew at most one meter away from the host plant and subsequently laid their eggs on stems,

dead plant material, rocks or other plants (randomly chosen by the female). In any case the support on

which the eggs were laid was never found to be higher than the host plant. In contrast, females of Oleria

onega agarista lay most of their eggs on the host plant. As animal and plant communities differ in

composition on the two sides of the mountain, we hypothesise that the distinctive behaviour of laying

eggs next to the host plant may be correlated with the presence of egg predators, as eggs might be

easier to find when laid on leaves of the host plant than when laid on other substrates.

The questions addressed here are:

Is female preference for S. mite positively correlated with larval preference and performance?

Do O. o. ssp. females lay their eggs in the vicinity of the host plant, instead of on them, to avoid

predation?

Do egg survival rates differ between the two Oleria subspecies?

Do larval activities during the day (and night) differ between the two subspecies in ways that

might result in avoidance of predation events?

How different is the soil surface fauna, and therefore the potential guild of natural enemies,

between the NE and SW slopes of the mountain?

Are some natural enemies present on the SW but not on the NE slope?

As the biology of Oleria remains unknown, the different developmental stages will be described

and the duration of each detailed

44



Study organisms. Oleria onega agarista (C. Felder and R. Felder) and Oleria onega ssp. nov.

are two closely related subspecies endemic to north-eastern Peru. O. o. ssp differs morphologically

from O. o. agarista by the narrower black edge of the wings and by to black bands on the hindwing that

are never connected (Lamas, pers. comm.). The subspecies agarista is characterised by a transversal

band connecting these two bands on their middle part on the Cu1 and Cu2 veins. The morphological

hybrids are recognisable by an incomplete transversal band. The butterflies used in this study came

from populations at the Shilcayo and Ahuashiyacu sites for O. o. ssp., and from Km30 in case of O. o.

agarista. The potential host plants that we tested here all belong to Solanaceae family and the section

Pteroidea of the genus Solanum (Knapp, 1997). Three of them, Solanum mite, Solanum anceps and

Solanum angustialatum are single-stemmed herbs that grow to a height of max. 1.5 m whereas the

fourth species, Solanum uleanum, is a small vine. The specimens of S. anceps and S. uleanum used

for feeding and for oviposition experiments in cages came from the Km30 site, S. angustialatum was

collected near the Tunnel (near the top of the mountain), and S. mite at Shilcayo.

Study sites: All fieldwork was carried in the north-eastern Peru, on the flanks of the mountain

"Cerro Escalera". The study sites are situated between the town of Tarapoto (SW slope) and Km30 of

the road (Carretera marginal) to Yurimaguas (NE slopes) (S 06°24- 06°28 and W 76°18 – 76°22). The

road crosses the Cerro Escalera mountain through a tunnel (alt. ≈ 1100 m). The two sides of the

mountain contrast in their climatic conditions: the SW side is usually sunny and hot (mean temp. ca.

30°C), while on the other side the climate is wet and rainy including fog and cooler temperatures (mean

temp. ca. 27° C). The Shilcayo and Ahuashiyacu sites are located on the SW slope, whereas the Km30

site is situated on the NE side.

For laboratory experiment, the butterflies were kept in cages made of wood and wire-mesh

(plastic and metallic) that was stapled to the wood. To reduce mortality due to ants, the cages were

standing raised above the ground. Cage size was 1m3, allowing the butterflies to fly. Constant access to

small recipients with 10% sucrose solution was insured in the cage. Heliotropium flowers were regularly

put into the cage as a source of pyrrolizidine alkaloids (PA’s) (used for chemical defence and as

pheromons precursors) as well as bird droppings as an amino acid source for egg production.

45


Immature development larval preference and performance on different host plant species.

Five to fifteen individuals of each host plant species were collected and potted to facilitate their

manipulation. Females of O. o. ssp. and O. o. agarista were separated into two cages, with 10-15

females per cage. One individual of one of the four plant species (S. mite, S. anceps, S. angustialatum

and S. uleanum ) was placed in each corner of the cage. Plants were exchanged or rotated daily in

cages, so that each day females were offered a new situation. Every second day all eggs laid were

collected and it was noted on which host plant they were laid. Eggs encountered on other substrates

within 40 cm from a plant were considered as associated with this plant.

A total of 120 individuals were reared from the egg to the adult stage on Solanum mite. The

duration of each stage was noted for both subspecies.

A total of 273 eggs laid in the cages by both subspecies were placed individually in the center of

Petri dishes with a piece of leaf of each of the four host plants around them. Just after hatching, the first

host plant that was fed upon by each larva was recorded and considered as preferred. We tested the

significance of correlation between larval preferences and female oviposition preference (see Chapter

1).

A further experiment was carried out with 130 eggs of both Oleria subspecies to compare larval

performance on the two most common host plants, S. mite and S. anceps. Each larva was fed from

hatching to pupation with the same plant species. Three parameters were taken into account to test the

suitability of the different plant species: larval mortality, duration of larval development and three

morphological measurements of pupal size (the length and two widths).

The survival of each subspecies on the two host plant was compared using Fisher's exact tests.

To compare the duration of larval and pupal development and the size of pupae of each subspecies on

the two plants, we used a general linear model ANOVA procedure for unbalanced data (Sokal & Rohlf,

1995).

Egg survival in the field

To estimate egg mortality rates due to natural enemies in the field and to compare these

between the two slopes of the mountain, a total of 400 eggs laid in cages were glued on plants of the

two main study places: Shilcayo (SW side) and Km30 (NE side). In each site 50 eggs of each

subspecies were glued on the underside of S. mite leaves and 50 others on stems of other plants,

46


stones, sticks, dead leaves etc. (50 O. o. ssp. + 50 O. o. agarista on leaves + 50 O. o. ssp. + 50 O. o.

agarista on other substrates x 2 sites = 400 eggs). As entomological glue is not resistant to rain we

used normal universal glue. To avoid direct contact of the eggs with the glue, a small part of the support

(plastic bags or on S. mite leaf) on which the egg has been laid was cut out. Eggs were exposed the

day they were laid. After two days in the field they were collected again to avoid their hatching during

the experiment. The number of eggs still present was counted and each egg was then put into a small

vial to study eventual parasitisation. At each site the whole experiment was performed within one week.

Differences in egg survival rate between the two subspecies in each site were analysed pairwise for the

same support (leaves or other substrates) by Fisher's exact tests. Differences between egg survival rate

on leaves and on other substrates were also analysed. Finally differences between the two sites (Km30

and Shilcayo) were compared. The eggs that had hatched were considered as having survival.

Succesfully hatched eggs are usually easy to recognise because the larvae almost always leave the

part of the eggshell glued by the female on the leaves, although the rest of the shell is eaten.

The second part of the egg survival experiment was carried out only at the Shilcayo site where

predation was expected to be higher. Only O. o. ssp. eggs were exposed. Fifty eggs were glued onto

each of fifty S. mite plants. Half of the plants were treated with protecting sticky glue around the stem,

thus "walking" predators were excluded. Eggs were re-collected and counted after two days. Results

were analysed with a Fisher's exact test between the two treatments.

Larval activity

A total of forty-one larvae of the two subspecies (O. o. ssp. from Ahuashiyacu and O. o. agarista

from Km30), were placed singly into vials to compare their circadian activity patterns. Larval activity may

have been affected by differential selection pressures related to predation. All larvae were reared from

the egg stage. Fresh leaves of S. mite were provided daily. During a period of 31 days larval activity

was checked four times daily: at 07.00h., 12.00h., 17.00h. and 22.00h.. Activities were classified as:

resting, walking, eating, moulting, and pupating. A factorial analysis was performed using the three main

activities: resting, walking, and eating, to compare the differences between the two subspecies. In a

second analysis, walking and eating were grouped, as being the activities during which larvae are more

exposed to natural enemies.

47


Predators

From October to February four malaise traps and ten Barber traps were installed once a month on each

mountain slopes. They were left for five days in the two main sites Km30 (NE slope) and Shilcayo (SW

slope). Arthropods collected were identified to family level for all of them, and additionally to genus level

for ants. The diversity of families (for insects) and order for (spiders and harvestmen), and arthropod

abundance, were compared between the two sites. Ants were collected manually at both sites to test

their potential role as egg predators in the laboratory. At the Shilcayo site where predation was

expected to be higher, ants were also collected during the night. A patch of twenty plants (host and non-

host) was checked during one hour and all the ants on them were collected with an entomological

aspirator and identified to genus level. At Km30 five hours of collection were performed because of the

less accessibility of the site, whereas at Shilcayo twenty hours were carried out during the day and

three hours during the night. Host plants with eggs on their leaves were potted and the pots were

sealed with paper lids to hinder ants from entering the soil in the pot. Ants were placed individually on

the paper lid and the behaviour of each ant was observed for 15 minutes. It was recorded whether each

ant climbed onto the plant, approached the eggs and whether any eggs were eaten. At least five

individuals of each ant species were tested, less abundant species with fewer than five individuals were

discarded. The behaviour of the ants when placed on the paper was noted. The species that showed an

aptitude to climb quickly on the plant were re-tested with at least twenty individuals.

RESULTS

Immature development, larval preference, and performance on different host plant species.

Eggs are small, oval and white, with small features on the surface. Eggs are always laid singly.

The egg stage lasts an average of three day. After hatching, the transparent larva first eats the eggshell

before feeding on the host plant leaves. Larvae become grey with one yellow stripers on each side of

the body. Oleria (and all Ithomiinae) have five larval instars, and at each moulting they eat the old skin

and leave only the sclerified head capsule; they moult every two or three days. After two weeks the

caterpillar stops eating, becomes green and begins pupation The time lapse between the final feeding

and pupation usually lasts about twelve hours, but in some cases the larvae can stay more than twenty-

48


four hours, perhaps waiting for appropriate conditions. The pupae is a beautiful green drop, with shiny

golden metallic reflects. There is some colour varieties (brownish or yellowish) in the same subspecies,

but this polymorphism did not affects the adult coloration. After 6 days the future wings become black

with white dots and the wing pattern of the imago can already be seen. After 7 or 8 days the butterfly

ecloses, drying its wings within one hour. Eclosion can take place at any time of the day.

Figure 1 shows the mean duration of each stage and the total duration of the larval (five instars)

and pupal stages. The results were obtained from 120 individuals from each of the two subspecies, fed

with S. mite. The durations of the larval stage differed slightly but significantly between O. o. ssp. and O.

o. agarista, with O. o. agarista showing a slightly shorter duration of larval development (Fig 2) (t-test p-

value= 0.003). Egg development took on average three days while larvae needed twelve days to finish

their development. The mean duration for the pupal stage was seven days. Standard errors were very

low, revealing little variation between individuals.

0

2

4

6

8

10

12

14

Egg st

age

Larva

l stag

e1

L. sta

ge2

L. sta

ge3

L. sta

ge4

L. sta

ge5

Total la

rval s

tage

pupa

l stag

e

Dev

elop

men

t tim

e (d

ays)

Fig. 1 : Results from observation of growth of immature stages of 120 butterflies of the two Oleria subspecies,

reared on Solanum mite.

Bars indicate the duration of the different stages : the eggs, the five larval stages separately, the whole larval stage,

and the pupae.

49


0

2

4

6

8

10

12

14

O.o.agarista O.o.ssp

Dev

elpo

men

t tim

e (D

ays)

EggLarvaePupae

Fig. 2 : Comparison of duration of development time between O. o. ssp and O. o. agarista.

Bars indicate the three different stages of the two subspecies. Larval development time was significantly longer for

O. o. ssp. (T-test : p= eggs 0.78, larvae 0.0038, pupae 0.666).

The correlation between female oviposition preferences and larval feeding preferences is shown

in figure 3. A total of 273 eggs were laid by the females of Oleria onega ssp. and agarista on the four

host plant. In both butterflies subspecies no differences were found between preferences of females

and larvae (result not reported here). Therefore data of the two subspecies are combined in this

analysis. Results revealed that S. mite is the host plant most preferred by both the females and the

larvae followed by S. anceps. The two species S. angustialatum and S. uleanum may be considered as

hosts but attraction to both plants is relatively low.

0

20

40

60

80

100

Pro

porti

on o

f cho

ice

(%)

Female choiceLarvae choice

S. miteS. uleanum

S. angust.S. anceps

Figure 3 : Comparison of female and larval host plant preferences (N= 273 eggs).

Females and larvae were offered four host plants, S. mite, S. anceps, S. angustialatum and S. uleanum. For both

female and larvae, data for both subspecies were analysed together, as no differences were observed between

them.

50


Larval performance was tested on the two most preferred host plants, S. mite and S. anceps,

with a total of 130 eggs from both subspecies (116 O. o. ssp. + 14 O. o. agarista). High mortality rates

were observed as only 65 individuals reached the pupal stage. The individuals that have reached the

pupal stage give the survival percentages (table 1) No differences in survival rates between the two

subspecies were found for either two host plants (Fisher's exact test: p > 0.4 for both plants).

Differences of survival between the two host plants were significant for O. o. ssp. with a better survival

on S. mite (Fisher's exact test: p = 0.0028) but not for O. o. agarista (Fisher's exact test: p > 0.5),

Analysis of variance (ANOVA) of larval development time of each butterfly subspecies on the two host

plants showed that there was no significant effect of the butterfly subspecies nor of the plant species

(between butterflies: F = 0.21, d.f. = 1, P > 0.5; between plants: F = 1.14, d.f. = 1, P > 0.2). The pupal,

development time was slightly longer for larvae that have been fed with S. anceps (between plants: F =

3.99, d.f. = 1, P =0.04), and the two subspecies did not differ in this aspect (between butterflies: F =

0.09, d.f. = 1, P > 0.5) (Table 1). Length of pupal was significantly different between the two subspecies

when reared on S. mite (F = 16.67, d.f. = 1, P = 0.000). However, a host plant effect was also

observable since pupae of both Oleria subspecies were smaller when reared on S. anceps (F = 63, d.f.

= 1, P = 0.000).

Table 1 : results of the development comparison of the two Oleria subspecies on two host plants : S. mite

and S. anceps. Four parameters were taken into account: the larval development time (larvdevt), the pupal

devevlopment time (pupdevt); the pupal length (Pupal L) and the proportion of individuals that reach adult stage

(survival).

S. mite S.anceps

larvdevt (days) pupdevt (days)

Pupal L. (cm) Survival (%)

larvdevt (days) pupdevt (days)

Pupal L. (cm) Survival (%)

O. o. ssp. 17.26 (± 0.21) 7.08 (± 0.10) 10.89 (± 0. 07) 67 (± 0.03) 16.95 (± 0.41) 7.32 (± 0.10) 10.01 (± 0.11) 38 (± 0.03)

O. o. agarista 17.33 (± 0.88) 6.83 (± 0.40) 10.29 (± 0.11) 86 (± 0.05) 16.25 (± 0.25) 7.50 (± 0.28) 9.44 (± 0.21) 57 (± 0.09)

Larvdevt = larval devevlopment time; pupdevt = pupal devevlopment time; Pupal L. = pupal length; survival= percent of individuals that reach adult stage

51


Eggs survival in the field

For this experiment 100 eggs of each butterfly subspecies were glued onto host plants (half of

the eggs) and other objects (half of the eggs) on the SW slope (Shilcayo) and the same was done on

the NE slope (Km30). No significant differences were noted between both subspecies with respect to

egg survival. (Fisher's exact test : p> 0.5 for all), therefore the results for the two subspecies were

pooled for the following analyses. Differences between survival rates on leaves and on other substrates

were analysed for each site. In Shilcayo a significantly lower survival rate was found for eggs laid on

host plant leaves compared with eggs laid on other substrates (Fisher's exact test: p = 0.000) (Fig.4A).

On the other hand, no differences in egg survival were noted between eggs laid on host plant leaves

and those laid on other substrates at the Km30 site (Fisher's exact test: p > 0.3) (Fig.4B). When

comparing the survival of eggs of each subspecies exposed on the same substrate but at different sites

it became evident that more eggs on leaves were missing in Shilcayo than at Km30 (Fisher's exact test:

p = 0.02). No differences in survival rate were found between the two sites for those eggs laid on

objects (Fisher's exact test: p > 0.3)

0

10

20

30

40

50

60

Leaves Sticks

Support

Num

ber o

f egg

s

PresentMissing

0

10

20

30

40

50

60

70

80

Leaves SticksSupport

Num

ber o

f egg

s

PresentMissing

(B)(A)

Fig 4 : Survivorship of eggs of two Oleria subspecies on different substrates in two sites : (A) Shilcayo (SW

slope) and (B) Km30 (NE slope).

Bars indicate the number of eggs present and absent after two days glued on leaves or on other objects. Data of

both Oleria subspecies were analysed together, as no significant differences were observed between them. Egg

survival was significantly higher when laid on other substrates in the Shilcayo site ( Fisher's exact test: : p= 0.000),

whereas no differences in survivorship were observed between eggs laid on leaves and on other substrates in the

Km30 site (Fisher's exact test: p> 0.3).

52


On plants protected by sticky glue (figure 5) egg survival was significantly greater than on non-

protected plants (Fisher's exact test: p= 0.022).

None of the eggs that were re-collected and placed in vials after the experiment were infested by

parasitoids.

01020304050

60708090

100

non-protectedplant

protected plant

Prop

ortio

n of

egg

s (%

)Presentabsent

Fig 5 : Survivorship of O. o. ssp eggs glued on Solanum mite leaves in Shilcayo, with and without sticky glue

on the stem of the host plant.

Bars indicate the proportion of eggs present and absent after two days in the field. Differences between the two

treatements were significant (Fisher's exact test: : p= 0.02)

Larval activity

Preliminary interpretation of the data revealed that only three activities were frequent: resting,

eating and walking. Larval activity was estimated for each subspecies (Figure 6). Our results revealed

that O. o. agarista larvae move around more than do larvae of O. o. ssp. At 7.00 a.m. both subspecies

showed little activity (resting). Between 17.00 and 22.00 larvae of O. o. ssp. showed a tendency to feed

more, whereas at that time O. o. agarista larvae are mostly walking. As walking and eating were both

activities where larvae are exposed to a higher predation risk, these two behavioural elements were

grouped to form the category “activity”. A second factorial analysis was performed with the categories

“rest” and “activity” (Figure 7). O. o. ssp. larvae showed a tendency to rest more than O. o. ssp. larvae.

In general, both tended to rest more at 7.00h than at other time. At noon O. o. agarista larvae were

more active than O. o. ssp. larvae. However, in the afternoon and at night O. o. agarista larvae reduced

their activity, whereas the contrary was the case for O. o. ssp. larvae.

53


-0,4

-0,3

-0,2

-0,1

0

0,1

0,2

0,3

0,4

-0,4 -0,3 -0,2 -0,1 0 0,1 0,2 0,3

O. o. agarista 7.00h

O. o. ssp. 7.00h

resting




walking

O. o. ssp . 22.00h

O. o. ssp. 17.00h eating

O. o. ssp . 12.00h

Figure 6 : Principal component analysis of the three main activities of the larvae of the two subspecies : O.

o. ssp. and O. o. agarista

Eating, walking and resting were the three main activities. Squares indicate the positions of the activities and the

positions of the different subspecies at four time : 7.00h, 12.00h, 17.00h, 22.00h.

0

1

2

3

4

5

6

7

8

9

-0,80 -0,60 -0,40 -0,20 0,00 0,20 0,40 0,60 0,80

Factor (mean)

activity restO. o. ssp7.00h

12.00h

17.00h

22.00hO. o. agarista7.00h

12.00h

17.00h

22.00h

Figure 7 : Principal component analysis of the two larval states, active and resting, of the two subspecies :

O. o. ssp and O. o. agarista

Activity and resting are the two states that occur in larvae. Black symbols indicate the positions of the O. o.

agarista larvaes between the two states and the empty symbols those of O. o. ssp. at four different hours during

the day: 7.00h, 12.00h, 17.00h, 22.00h.

54


Predators

Qualitative and quantitative differences in the composition of the fauna of the potential predator

between the two slopes were assessed in five different months (Table 2). Ants and Gryllidae were much

more abundant in Shilcayo (SW slope) than at Km30 (NE slope).

Table 2 : list of the potential predator groups caught with "Barber" and "Malaise" traps, in the two sites :

Shilcayo (SW slope) and Km30 (NE slope)

Shilcayo Km30 Malaise traps Number of individuals caught Number of individuals caught

oct nov dec jan feb oct nov dec jan feb Araneae 0 1 1 1 2 Blattodea 3 1 0 0 0 Blattodea 4 3 3 0 1 Gryllidae 1 1 0 0 0 Gryllidae 3 1 0 0 1 Formicidae 12 10 2 4 5 Reduviidae 0 0 4 0 0 Staphylinidae 1 0 1 0 0 Formicidae 3 15 83 8 11 Cicindelidae 0 0 1 0 0 Carabidae 0 0 1 0 2

Barber traps Number of individuals caught Number of individuals caught

oct nov dec jan feb oct nov dec jan feb Araneae 2 5 14 16 7 Araneae 4 3 6 4 3 Opiliones 1 1 1 6 1 Opiliones 4 6 3 0 5 Blattodea 9 3 18 2 9 Blattodea 2 2 1 1 9 Gryllidae 16 24 66 67 58 Gryllidae 12 9 2 3 9 Formicidae 25 76 538 126 224 Formicidae 67 78 23 73 99 Staphylinidae 3 1 18 5 7 Staphylinidae 6 4 5 1 4 Histeridae 3 0 2 0 1 Histeridae 0 1 1 1 1 Reduvidae 2 0 2 0 1

Three genera of the predatory ants of subfamily ponerinea were found in Shilcayo but not at

Km30 (Table 3). These were Ectatomma, Hypoponera and Odontomachus. However, only few

individuals of the latter two genera were found. At Km 30, another ponerinea genus, Prionopelta, was

found that does not occur in Shilcayo, but only one individual was caught. However three genera of the

predatory subfamily Ecitoninae (Eciton, Labidus and Neivamyrmex) were recorded in greater

abundance in Km30.

55


Table 3 : Ant genera caught with Barber and Malaise traps in the two sites : Shilcayo (SW slope)

and Km30 (NE slope). Nb sp = number of species per genus, and food = alimentation. habits : p = predators,

o = omnivorous; n= nectaries; c= fungi

Shilcayo Km30 Genus Nb sp. Food Genus Nb sp. Food

Formicinae Ecitoninae Camponotus 4 n Eciton 1 p Paratrechina 2 p Labidus 1 p Myrmicinae Neivamyrmex 1 p

Crematogaster 2 o Formicinae

Acromyrmex 1 c Camponotus 1 n Cephalotes 1 n Myrmelachista 1 n Mycocepurus 1 c Paratrechinae 1 p Pheidole 5 o Myrmicinae

Solenopsis 2 o Cephalotes 1 n Ponerinae Megalomyrmex 1 o

Ectatomma 2 p Monomorium 1 o Hypoponera 1 p Ochetomyrmex 1 o Odontomachus 1 p Pheidole 3 o Pachycondyla 3 p Solenopsis 1 o Wasmannia 1 o Ponerinae

Pachycondyla 3 p Prionopelta 1 p

In Shilcayo, fifteen ant species were manually collected in the field using an aspirator. Their

aptitudes to climb on plants and their interest in eggs were tested in the laboratory. At Km30 we

collected twenty-three species, but only nine were abundant (Table 4). Among the sixteen species

caught in Shilcayo (Table 4) six climbed on the plants. Five species belonged to the predator genus

Ectatomma, and one to the omnivorous genus Crematogaster. For each of the six species their interest

and their ability to eat the eggs were tested in the laboratory with twenty to thirty individuals (depending

on the availability of the species). In these experiment, none of these species ate eggs of Oleria and

their interest in eggs was generally low. Odontomachus sp. was not tested as only four individuals were

caught and only at night. No individual of Hypoponera sp. was caught. Ectatomma species were more

active during the day, but were also easily captured at night, as they rest on plants. This genus was

present only on the SW slope. The ants of the NE slopes were only tested with five to ten individuals

56


because of the few availability of ants. No further observations were done on them, as no special

oviposition behaviour was observed there on Oleria onega agarista. Genera including potential predator

on the NE slope were mostly Eciton and Camponotus. Camponotus species were also found at Km30.

Eciton species are army ants that attack other ant colonies, but also eat other arthropods.

Table 4 : Observation of ant behaviour when artificially placed near a host plant with eggs: food habits,

their aptitude to climb, their reaction if put on the leaves (stay or go down), the estimation of the main behaviour

(running, leaves inspection or grooming) their interest in Oleria eggs and an estimation of their density. Symbols

for all the columns excluding density : - = not observed; + observed; ++ = observed very frequently. Symbols for

density: - = 5 individuals caught, + = 5-10 individuals caught; ++ = 10-40 individuals caught.

Alim. climb on the plant

stay if put on the plant

Running Inspect. of the leaves

grooming taste eggs

High density in the field

Shilcayo Formicinae Camponotus sp.A n - + - + + - ++ Camponotus sp.B n - + + + + - - Formica sp.A o - + - + + - + Myrmicinae

Cardyocondila o - - + - - - - Crematogaster sp.A o + + + - - - ++ Dendromyrmex o - + + + - - + Eucryptocerus o - - + - - - + Pheidole sp.A o - - + - - - ++ Pheidole sp.B o - + + + - - ++ Ponerinae Ectatomma sp.A p ++ + - ++ + - ++ Ectatomma sp.B p - + - + + - + Ectatomma sp.C p + + - + + - + Ectatomma sp.D p + + - + ++ - + Ectatomma tuberculatum p + + + + - - + Ectatomma sp p + + - + + - ++ Odontomachus p - + + - + - -

Km30 Ecitoninae Neivamyrmex o + + + + - + - Formicinae Formica sp.B o - + - + + - + Formica (Camponotus?) o - + - + + - ++ Myrmicinae Crematogaster sp.B o + + - + + - + Crematogaster sp.C o + + + - - - + Megalomyrmex o - - + - + - - Pheidole sp.C o - - - + + - + Pheidole sp.D o - - + - - - + Pseudomyrmicinae Pseudomyrmex o + + + - - - +

57


DISCUSSION AND CONCLUSION

Biology and performance of the larvae

The study on the duration of each immature stage has revealed that the egg stage lasted on

average three days. Either of the five larval stages varied between two or three days in length. The first

and fifth larval stages were slightly longer in time than stages two to four. The longest stage was

pupation, which lasted seven to eight days. In the first observations on immature development time,

duration of egg and pupal stages were not significantly different between the two subspecies. However,

duration of larval development was one day longer (fig. 2) in Oleria onega ssp. This result was

unexpected as we assumed that selection pressures by natural enemies was higher on the SW side, to

which O. o. ssp. is restricted. As a consequence, the development of O. o. ssp. should be as short as

possible to avoid the danger of being killed. However, in the experiment comparing performances on

the two host plants Solanum mite and S. anceps (Table 1), the larval stage was longer than in the first

observation on immature stage development (Figure 1). This was probably due to more favourable

temperature conditions in the second experiment (humidity and temperature). Therefore, a realistic

estimate of the duration of the larval stage in the field is probably between twelve and seventeen days,

depending on temperature and humidity.

Among the four host plants tested, S. mite was the one most preferred by females in terms of

oviposition choice (Chapter 1) as well as by larvae in terms of food choice. This was true for both Oleria

subspecies, demonstrating that female and larval choices were highly correlated. Also, larval

performance has showed positive by correlated with these preferences. Larval development time did

not differ significantly between the two host plants. Pupal development was prolonged when larvae fed

on S. anceps. Significantly larger pupal length and higher larval survival rate both showed that S. mite is

a better host than S. anceps for both subspecies. The widths of the pupae were not taken into account

as they revealed only the water content of the pupae. The differences in length are probably due to the

fact that larvae eat less on S. anceps, leaves of which are tougher than those of S. mite. Also, S.

anceps may be less digestible for the larvae than is S. mite. To define better the differences in the

suitability of the four potential host plants to the larvae, analyses of their chemical compounds would be

helpful. Such studies would show if S. mite is richer in nutrient, or perhaps possesses chemicals, which

are more attractive for both larvae and females.

58


Relationship between biology, performance and predation

We discussed in the first part of this chapter that larval development was slightly longer for O. o.

ssp. than for O. o. agarista. This fact is in conflict with the notion that larval development should be as

short as possible at those sites were predation appears to be comparatively high. In this part we will

discuss how larval activity is related to larval survival.

We first compared the three main activities, resting, eating, and walking, between the two

subspecies at different times of day. Thereafter, eating and walking were grouped together as activities

during which larvae are more exposed to natural enemies in the field. The comparison showed that O.

o. ssp. spent more time resting than did O. o. agarista. This mean to stay more hidden. than the other.

Both subspecies tended to rest more in the morning from 7.00 hours till noon than at other times of day.

O. o. ssp. fed during the afternoon hours whereas O. o. agarista (= Km30) fed mostly at midday. The

main activities of O. o. agarista during the afternoon were feeding and walking. The observation that O.

o. agarista larvae move more than these of O. o. ssp. can be related firstly to the fact that this

subspecies suffers less predation than the other (Machado & Freitas, 2001) and secondly to the

observation that S. mite is less abundant on the NE slope. This implies that larvae would need to

wander around more to find their less abundant host plants. O. o. ssp. on the other hand is less active:

it rests in the morning and feeds from midday on. Larvae usually rest on the underside of the host plant

leaves, at the base of the stem, or on other objects near the host trunk.

The field experiment in which eggs were placed either on the host plant or on objects next to the

plant showed that on the SW slope eggs had a higher rate of survival when placed on objects near the

host plant rather on the plant. On this slope, when plants were protected by glue, egg disappearance

was significantly higher on non-protected plant. This observation and the fact that O. o. ssp. larvae are

less mobile than those of O. o. agarista led us to the conclusion that this subspecies could be more

exposed to natural enemies and thus requires protection of eggs from predators that walk on the

leaves. Also, the larvae seem to avoid being exposed on the leaves for too long and therefore reduce

the risk of possible predation. This fact will be discussed here in terms of predation and not of natural

enemies, as parasitism was never observed in the experiments. As these behaviours were observed

only on the SW slope, they are probably due to one or several predator groups that climb on the host

plants and do not occur on the NE side. The one predator group that is abundant, climbs on the plants,

and patrols their leaves, are certain ant species. However, the utilisation of Barber traps in ant capture

59


may be discussed as the quantity of the different genera of ants captured may vary drastically

depending on ethological and physiological factors of the insects (Seifert, 1990). When manually

collected in the field, the only abundant predatory genus that was exclusively found on the SW side but

not on the NE side was the Ponerinea genus Ectatomma. Ectatomma are medium-sized ants (~ 1 cm.)

which showed careful observation behaviour of the leaves and after 5 to 10 min of exploration started

grooming. Ants of the genus Ectatomma may be categorised as "calm" ants in comparison to ants of

genera such as Pheidole and Camponotus, which were found to show very nervous activity when

manipulated. In other genera observed, grooming was the first activity done by the ant after being laid

on the paper dish that sealed the pot of the plant, as for example Camponotus, Pheidole and

Crematogaster usually first run. The aptitude to climb on the plant and to search food on them is more

easily assessed for “calm” ants than for “hectic” ones, as in this latter case the fact of climbing may be

completely aleatory.

Relationship between ants and butterflies has been studied mostly in terms of predation,

However cases on mutualism or associations has been reported. Previous works have revealed

different types of antagonistic relationships between ants and herbivores: ants may prey on immature

satges of butterfly (Letourneau, 1983; Heads & Lawton, 1985; Costa et al., 1992; Freitas & Oliveira,

1996) herbivore eggs may be removed by ants and dropped to the ground, resulting in enhanced egg

mortality (Letourneau, 1983), or plants may be occupied by ants, which can affect the oviposition

behaviour of the butterflies (Inouye & Taylor, 1979; Freitas & Oliveira, 1996). The presence of ants on

plants is often due to the production of nectar in extrafloral nectaries (EFN's) by the plant (Turlings &

Wäckers, 2002). On the other hand, obligatory mutualism has been found in many Lycaenid species

(Atsatt, 1981a, b; Pierce & Edgar, 1985) for which ants are necessary for larval development. Cases of

facultative mutualism was reported in riodinids where ants defend the butterfly larva on their host plant

against predators (DeVries, 1991). An intermediate case between predation and mutualism was found

in an Ectatomma ant species and the Ithomiinae genus Mechanitis on a Solanum species. The ants

profit from the damage done by the young larvae on the leaves by consuming the exsudates (Young,

1978c). In this special case, neither predation by the ants on the larvae was noted nor did the ants offer

protection against predators. Published studies of Ectatomma have mostly been related with their habit

of feeding on EFN's (Costa et al., 1992; Morrone et al., 2000; Bluethgen et al., 2000; Apple & Feener,

2001), and they are considered as potential predators. However no special predation events were

60


recorded by this genus on herbivores as it is the case with other genus like Myrmica, Formica (Heads &

Lawton, 1985) Camponotus (Oliveira et al., 1987; Machado & Freitas, 2001) and Pheidole (Letourneau,

1983; Machado & Freitas, 2001). Even so, studies of Ectatomma ruidum have revealed that immature

stage lepidoptera may represent up to 5.4% of their alimentation (Ibarra-Núñez et al., 2001). We cannot

assert that Ectatomma are the ultimate cause of O. o. ssp. laying eggs next to the host plant, as no

predation act was observed when eggs were offered to the ants. However, they are considered as

predators in the literature (Schatz & Wcislo, 1999; Schatz et al. 1999) Furthermore, we observed that

Ectatomma individuals usually spend most of their time on plants (Solanum spp. and others). And

importantly, they are present and common only on the SW side. Therefore, we suggest that they may

constitute some of selection pressure, either as egg predators, or in repelling ovipositing females

leading to the behaviour laying eggs close to the host plant rather than on the host plant itself.

ACKNOWLEDGEMENTS

This work was supported by a Grant from the Commission for travel of the Swiss Society of Natural

Science (SANW), and the Matthey-Wüthrich funds for travels. Thanks are due to Dr. G. Lamas

(Universidad Mayor de San Marcos, Lima) for determination of Ithomiinae, to Dr. J. Delabie

(CEPEC/CEPLAC, Itabuna, Brasil) and to Dr. D. Chérix (University of Lausanne, Switzerland) for

determination of ant. To Dr M. Joron, (University of Leiden, Netherland) and Dr. Karl Gotthard

(University of Neuchâtel), for their helpful comments on the manuscript, and to Jacqueline Moret

(University of Neuchâtel) for statistical support. Thanks are due to G. Valencia, M. Abanto, and J.

Reategui for field collaboration.

61


REFERENCES

Apple, J.L.; Feener, D.H.Jr. 2001. Ant visitation of extrafloral nectaries of Passiflora: the effects of

nectary attributes and ant behavior on patterns in facultative ant-plant mutualism. Oecologia 127 (3):

409-416.

Atstatt, P.R. 1981a. Ant-dependent food plant selection by the mistletoe butterfly Ogyris amaryllis

(Lycaenidae). Oecologia 48: 60 - 63.

Atstatt, P.R. 1981b. Lyceanid butterflies and ants: selection for enemy - free space. American

Naturalist 188: 638 - 654.

Barker, A. M.; Maczka, C.J.M. 1996. The relationships between host selection and subsequent larval

performance in three free-living graminivorous sawflies. Ecological Entomology 21: 317-327.

Bates, H.W. 1862. Contribution of the insect fauna of Amazon Valley. Transactions of the Linnean

Society 23: 495.

Bluethgen, N.; Verhaagh, M.; Goitia, W.; Jaffe, K.; Morawetz, W.; Barthlott, W. 2000. How plants

shape the ant community in the Amazonian rainforest canopy: the key role of extrafloral nectaries and

homopteran honeydew. Oecologia 125 (2): 229-240.

Brower, L.P. 1984. Chemical defence in butterflies. In: Ackery RIV-WaPR, ed. The biology of

butterflies. London: Academic Press.

Brown, K.S. 1979. Ecologia Geográfica e Evolução nas Florestas

Neotropicais (2 vol.). Campinas, Brazil.: Universidade Estadual de Campinas.


predator. Nature 309 (5970): 707-709.



Brown, K.S. 1987. Chemistry at the Solanaceae/Ithomiinae interface. Ann. Missouri Botanical Garden

74: 341-350.

Brown, K.S.; Freitas, A.V. 1994. Juvenile stages of Ithomiinae: overview and systematics

(Lepidoptera: Nymphalidae). Tropical Lepidoptera 5 (1).

Brown, K.S.; Vasconcellos, N.J. 1976. Predation on aposematic Ithomiine Butterflies by Tanagers


Chew, F. S. ; Robbins, R. K. 1984. Egg-laying in Butterflies. In: Van Wright R, Ackery P. R., ed. The

62


biology of Butterflies. New York: New York Academic. 65-79.

Clancy, K.M.; Price, P.W.; Sacchi, C.F. 1993. Is leaf size important for a leaf-galling sawfly (

Hymenoptera, Tenthredinidae)? Environmental Entomology 22: 116-126.

Costa, F.M.C.B.; Oliveira-filho, A.T. 1992. The role of extrafloral nectaries in Qualea grandiflora

(Vochysiaceae) in limiting herbivory: an experiment of ant protection in cerrado vegetation. Ecological


Craig, T.H.; Itami, J.K.; Shantz, C.; Abrahamson, W.G., Horner, J.D.; Craig J.V. 2000. The influence

of host plant variation and intraspecific competition on oviposition preference and offspring performance

in the host races of Eurosta solidaginis. Ecological Entomology 2000: 7-18.

DeVries, P.J. 1991. Mutualism between Thisbee irenea butterflies and ants, and the role of ant ecology

in the evolution of larval-ant associations. Biological Journal of the Linnean Society 43: 179-195.



Edmunds, M. 1974. Defence in Animals. A survey of anti-predator defences. Longman group, Harlow,

U.K.

Freitas, A.V.S.; Oliveira, P.S. 1996. Ants as selective agents on herbivore biology : effects on the

behaviour of a non-myrmecophilous butterfly. Journal of Animal Ecology 65: 205-210.

Frost, S. W. 1959. Insect life and natural history. New-York.

Haber, W.A. 1978. Evolutionary Ecology of Tropical Mimetic butterflies: Function of Ithomiine scent

scales in aggregating behaviour: PhD thesis, University of Minnesota.

Harris, M.O.; Sandanayaka, M.; Griffin,W. 2001. Oviposition preferences of the Hessian fly and their

consequences for the survival and reproductive potential of offspring. Ecological Entomology 26: 473-

486.

Heads, P.A.; Lawton, J.H. 1985. Bracken, ants and extrafloral nectaries. III How insect herbivores

avoid ant predation. Ecological Entomology 10: 29-42.

Higashiura, Y. 1989. Survival of eggs in the gypsy moth Lymantria dispar II. Oviposition site selection

in changing environments. Journal of Animal Ecology 58: 413-426.

Ibarra-Nunez, G.; Garcia, J.A.; Lopez, J.A.; and Lachaud, J.P. 2001. Prey analysis in the diet of

some ponerine ants (Hymenoptera : Formicidae) and web-building spiders (Araneae) in coffee

plantations in Chiapas, Mexico. Sociobiology 37: 723-755.

63


Inouye, D. W.; Taylor, O.R. 1979. A temperate region plant-ant-seed predator system : consequences

of extrafloral nectar secretion by Helianthella quiquenervis. Ecology 60 (1): 1-7.

Joron, M. 2000. Coloration avertissante et mimetisme müllérien: le problème de la diversification.

Thèse de doctorat, Université des Sciences et Technique du Languedoc. Montpellier: Académie de

Montpellier.



Knapp, S. ; Helgason, T. 1997. A revision of Solanum section Pteroidea: Solanaceae. Bull. nat. Hist.

Mus. London 27, no1: 31 - 73.

Letourneau, D.K. 1983. Passive aggression: An alternative hypothesis for the Piper-Pheidole

association. Oecologia 60: 122-126.

Machado, G.; Freitas, A.V.L. 2001. Larval defence against ant predation in the butterfly Smyrna

blomfildia. Ecological Entomology 26: 436-439.

Mallet, J.L.B. 1989. The genetics of warning color in peruvian hybrid zones of heliconius- erato and

heliconius-melpomene. Proc. R. Soc. Lond. B. Biol. Sci. 236: 163-185.

Mallet, J.L.B. 1993. Speciation raciation and color pattern evolution in heliconius butterflies evidence

from hybrid zones. In: Harrisson, R.G. ed. Hybrid zones and the Evolutionary process. New - York:

Oxford university press, inc.pp. 226-260.

Morrone, O.; Vega, A.S.; Maier, M. 2000. Elaiosomes in Urochloa paucispicata (Poaceae:

Panicoideae: Paniceae): Anatomy and chemical composition. Flora (Jena) 195 (4): 303-310.

Müller, F. 1876. Ituna and Thyridia: a remarkable case of mimicry in Butterflies. Kosmos: 100-109.

Nylin, S. ; Janz, N. 1993. Oviposition preference and larval performance in Polygonia c-album


Oliveira, P.S.; da Silva, A.F.; Martins, A.B. 1987. Ant foraging on extrafloral nectaries of Qualea

grandiflora (Vochysiaceae) in cerrado vegetation: ants as potential antiherbivore agents. Oecologia 74:

228-230.

Pierce, N.E.; Elgar, M.A. 1985. The influence of ants on host selection by Jalmenus evagoras, a

myrmecophilous lyceanid butterfly. Behavioural Ecology and Sociobiology 16: 209-222.

Pires, C.S.S., Price, P.W., Fontes, E.G. 2000. Preference-performance linkage in the neotropical

spittlbug Deois flavopicta and its relation to the Phylogenetic Constraints Hypothesis. Ecological

64



Rausher, M. 1979. Larval habitat suitability and oviposition preference in three related butterflies.

Ecology 60 (3): 503-511.


Waiblingen.

Schatz, B.; Lachaud, J.P.; Beugnon, G.; Dejean, A. 1999. Prey density and polyethism within hunting

workers in the neotropical ponerine ant Ectatomma ruidum (Hymenoptera, Formicidae). Sociobiology

34: 605-617.

Schatz, B.; Wcislo, WT. 1999. Ambush predation by the ponerine ant Ectatomma ruidum Roger

(Formicidae) on a sweat bee Lasioglossum umbripenne (Halictidae), in Panama. Journal of Insect

Behavior 12: 641-663.

Seifert, B. 1990. Wie wissenschaftlich wertlose Fangzahlen entstehen- Auswirkungen artspezifischen

Verhaltens von Ameisen an Barberfallen direckt beobachtet. Entomologische Nachrichten und Berichte

34: 21-27.

Singer, M.C.; Lee, J.R. 2000. Discrimination within and between host species by a butterfly:

implications for design of preference experiments. Ecology Letters 3: 101-105.

Sokal, R.R.; Rolf, F.J. 1995. Biometry : the principles and practice of statistics in biological research,

3rd edition. W.H. Freeman and Co eds. New-York.

Turlings, T.C.J ; Wäckers, F. 2002. Recruitment of predators and parasitoids by herbivore-injured

plants In: Cardé RT, Millar J eds Advances in Insect Chemical Ecology. Cambridge University Press. In

press.



Williams, K.S. 1983. The Coevolution of Euphydryas chalcedona Butterflies and their larval host plants.

Oecologia 56: 336-340.

Young, A.M. 1972. On the life cycle and natural history of Hymenitis nero (Lepidoptera, Ithomiinae) in

Costa Rica. Psyche 79: 284-294.

Young, A.M. 1973. The life cycle of Dircenna relata ( Ithomiidae) in Costa Rica. Journal of the

Lepidopterists' Society 27: 258-267.

65


Young, A.M. 1974a. Note on the biology of Pteronymia notilla ( Ithomiidae) in Costa Rican mountain

forest. Journal of the Lepidopterists' Society 28: 257-268.

Young, A.M. 1974b. On the biology of Godyris zavaleta caesiopicta (Lepidoptera: Nymphalidae:

Ithomiinae). Entomological News 85: 227-238.

Young, A.M. 1974c. A natural historical account of Oleria zelica pagasa (Lepidoptera: Nymphalidae:

Ithomiinae) in a Costa Rican mountain rain forest. Studies on the Neotropical Fauna 9: 123-139.

Young, A.M. 1977. Note on the biology of Hypothyris euclea in Costa Rica (Lepidoptera, Ithomiinae) in

Costa Rica. Pan- Pacific Entomologist 53: 104-113.

Young, A.M. 1978a. Note on the biology of the butterfly Hypoleria cassotis (Bates) (Nymphalidae:

Ithomiinae) in northeastern Costa Rica. Brenesia 14/15: 97-108.

Young, A.M. 1978b. The biology of the butterfly Aeria eurimedea agna (Nymphalidae: Ithomiinae:

Oleriini) in Costa Rica. Journal of the Kansas Entomological Society 51: 1-10.

Young, A.M. 1978c. Possible evolution of mutualism between Mechanitis caterpillars and an ant in

northeastern Costa Rica. Biotropica 10 (1): 77-78.

66

Chapter 3

Genetic (RAPD) diversity within and between Oleria

onega agarista and Oleria onega ssp. (Ithomiinae,

Nymphalidae, Lepidoptera) in north-eastern Peru

Based on:

Gallusser, S.; Guadagnuolo, R.; Rahier, M.

Genetic (RAPD) diversity within and between Oleria onega agarista and Oleria onega ssp.

(Ithomiinae, Nymphalidae, Lepidoptera) in north-eastern Peru. In preparation for

submission to Molecular Ecology

Chapter 3 Molecular study of two Ithomiinae subspecies

INTRODUCTION

It is widely accepted that distributions of species arise as a product of three processes:

speciation, extinction and transformation (e.g. Gaston, 1998). As a general picture, species having a

larger distribution range, have also a greater probability to be divided by a barrier (climatic, geographic,

environmental or other), than do species having a small range, and are thus more susceptible to

speciation processes. The frontier where two partially interfertiles population or species meet (e.g. two

different mimicry forms of unpalatable prey taxa), is known as a "hybrid zone". Fundamentally, a hybrid

zone is a cline or set of clines between two parapatric or sympatric hybridising taxa (Hewitt, 1988), and

they are usually very stable, while contact zones are sites were two species meet without hybridisation.

Because of the strong selection on rare forms, known as frequency dependent selection, hybrid and

contact zones are generally narrow (5-10 km) (Barton & Hewitt, 1985 1989; Mallet, 1989; Jiggins et al.,

1996). Hybrid zones are maintained by two antagonistic processes: on one hand selection (Mallet &

Barton 1989), ecological factors and adaptations (Jiggins et al., 1996) which act against gene flow, and

on the other hand by dispersal that tend to lead to speciation.

In this study, we clarify the taxonomical status of two Ithomiinae subspecies and their

morphological hybrids. This study concentrated on two Ithomiinae subspecies: Oleria onega ssp. nov,

thus far undescribed (Lamas, pers. comm.), and O. o. agarista (C. Felder and R. Felder, 1862). They

inhabit the easternmost mountain chain of the Andes before the Amazonian plain, in north-eastern

Peru, near the small town of Tarapoto. This mountain chain, called the Cerro Escalera, is considered an

ecological barrier for various organisms, principally butterflies (Joron 2000, Joron et al. 2001; Shulte

1999; Mallet, 1989, 1993). On the Tarapoto side, where the climate is hotter, we found Oleria onega

ssp., while only O. o. agarista, the subspecies of the lowland forests, was found on the other slope

characterised by a wetter and cooler climate. The two subspecies cohabit in only two locations: Estero,

near Shapaja, and the Bocatoma of Ahuashiyacu, near Tarapoto (Chapter 1). In this latter site, both

subspecies are present but apparently do not hybridise, whereas in Estero morphological hybrids were

observed. Therefore, Ahuashiyacu could be considered as a "contact zone", while Shapaja could be an

"hybrid zone".

Oleria onega agarista (C. Felder and R. Felder, 1862).and Oleria onega ssp. are considered as

conspecific by Lamas (Lamas pers. comm.). However, the status of O. o. ssp., as a subspecies or a

67


completely different Oleria species, is not already established. O.o. ssp. differ morphologically from O.

o. agarista by the narrower black edge of the wings, and by two black bands on the hindwing that are

never connected (Lamas, pers. comm.), whereas in Oleria onega agarista a transversal band connect

these two bands on their middle part on the Cu1 and Cu2 veins. The morphological hybrids are

recognisable by an incomplete or absent transversal band. As hybrids are rare in the zone where they

sympatric, we can suppose that a biological barrier exists, preventing crossings. Therefore, in a

previous study (Chapter 1), the two subspecies were put together to interbreed in laboratory, in a

search for eventual pre- or post zygotic barriers. However, females born in cages never accepted

males, regardless of whether they were from the same subspecies or not, and regardless of whether

males were wild or born in cages. Some behavioural differences between O. o. ssp. and O. o. agarista

were also observed. Indeed, females of the first species lay eggs up to 1 meter away from the host

plant, whereas O. o. agarista lay them mostly on the leaves of the host-plant (Chapter 1).

To gain a better understanding of the genetic relationship between these two subspecies, it was

important to assess the range of their genetic variation. A suitable tool is the RAPD technique (Williams

et al, 1990) which provides a virtually unlimited number of neutral DNA markers (Williams et al, 1990)

and is therefore an appropriate method for initial, overall analysis of variation between populations. In

addition, material for RAPD can be collected in the field and stored in alcohol, avoiding the necessity for

freezing or immediate processing, required for other techniques like isozymes (Bartish et al. 1999).

Moreover, RAPD markers can easily detect differences among populations and species of different

organisms, including plants (Ayres et al., 1999; Bussell 1999, Bartish et al., 1999; Skotnicki et al., 1999;

Comes & Abbott, 2000; Guadagnuolo et al., 2001a, 2001b), invertebrates (Ritchie et al., 2001; Moya et

al., 2001) and vertebrates ( Clausing et al., 2000; Vucetich et al., 2001). Two major and often mentioned

drawbacks of RAPDs markers are the lack of reproducibility and the loss of complete genotypic

information, due to the fact that most RAPD bands are dominantly inherited. However, the problem of

non-reproducible fragments can be highly reduced by using only high-quality DNA and by careful

optimisation of the PCR conditions (Guadagnuolo et al. 2001c, Wiesing et al.1994). Moreover, an

Analysis of Molecular Variance (AMOVA), which is not influenced by the dominance of the used

markers, can be used to determine the partitioning of RAPD variation between and within populations

(Huff et al. 1993).

68


The aims of this study were: i.) to elucidate, at least partly, the taxonomical relationships

between the two subspecies by investigating the genetic similarity between them, ii.) to investigate the

molecular variation among populations and iii) to determine the status of the hybrids in relation to O. o.

ssp. and O. o. agarista.


Population sampled:

A total of 120 samples of Oleria onega ssp., O. o. agarista and putative hybrids between them

were collected from 7 populations near Tarapoto and on the Cerro Escalera (Figure 1 Chapter 1) (Table

1). In order to evaluate genetic distances between the two subspecies, we also collected seven

specimens of a different Oleria species, Oleria gunilla serdolis (Haensch, 1909), and five specimens of

a different Ithomiinae genus, Hyposcada zarepha flexibilis (Haensch, 1909).

Three different conservation methods were tested: (i) in a subset of samples, butterflies were dried and

conserved in well closed boxes with silicagel, (ii) for a second subset the butterflies were conserved in

70% alcohol, and (iii) 99% alcohol was used for the third subset.

69


Table 1 : Sampling localities of Oleria onega ssp., O. o. agarista, O. gunilla serdolis, and Hyposcada zarepha

flexibilis, number of individuals analysed and abbreviations used in the results of the cluster analysis presentedd.

Populations Nb of individuals Abbreviation

Oleria onega ssp

Shilcayo 15 S

Urahuasha 7 U

Ahuashiyacu 11 A

San Roque 13 SR

Shapaja 9 SH

Oleria onega agarista

Km30 18 K

Km28 5 KM

Ahuashiyacu 2 AA

O. o. ssp/agarista Hybrids

Shapaja 7 HSH

Oleria gunilla serdolis (SanRoque) 6 SRogs

Hyposcada zarepha flexibilis (S. Roque) 2 SHhz

Hyposcada zarepha flexibilis (Shapaja) 1 SRhz

DNA extraction

Samples conserved in alcohol were washed three times with deionized water, and then dried

with paper.

In order to obtain enough DNA for numerous RAPD reactions, the DNA extractions were

performed on abdomens, which contain a large number of cells and are consequently rich in DNA.

However, the ends of abdomens were reminded to avoid possible contamination by the spermatophore

of male origin in females, as well as the sclerified parts of male abdomens (Aubert et al., 1999;

Schneider et al., 1997). The abdomens were ground in 2 ml Eppendorf tubes containing liquid nitrogen.

Extractions were then performed using the QIAGEN DNeasy Kit for animal cells (QIAGEN Inc.)

according to the manufacturer's instructions and DNA was resuspended in TE (pH 8, Tris 10mM-EDTA

1mM). Because of degradation of the dried material, DNA extractions were finally performed only on the

133 butterflies conserved in alcohol. The integrity of DNA was tested on a 0.8% agarose gel, and only

96 out of the 133 extracted individuals were suitable for subsequent RAPD analyses. DNA was

70


quantified using a WPA lightwave S2000 spectrophotometer and samples were then diluted at a

concentration of 5 ng/µl and stored at -20°C.

RAPD markers

All the PCR reactions were carried out in a final volume of 25 µl containing 25 ng genomic DNA,

2.5 µl10x PCR buffer (with 1.5 mM MgCl2), 0.25 µL 10mM dNtp, 0.5 µl primers (10 pmol/µl), and 0.1 µl

Qiagen Taq polymerase (QIAGEN Inc). Amplifications were performed with a Biometra T-Gradient

thermocycler with the following profile; 3 min at 94°C, followed by 39 cycles at 94°C for 45 s, 40°C for

45 s, 72°C for 1 min. and a final extension step of 10 min at 72°C . PCR products were mixed with 1/5

vol. loading buffer and separated on a 1.2 % (w/v) agarose gel, containing 0.4 µg/ml ethidium bromide,

in 0.5 X TBE at 100 V for 1 hour. DNA fragments were then visualised under UV light.

We tested 23 random primers of the series OPB, OPP and OPT (Operon Technologies, Alameda

California) on seven samples of the two subspecies, and repeated amplifications three times. Negative

controls were added at each PCR. Seven of these primers, OPB-01, OPB-11, OPB-12, OPB-15, OPT-

01, OPT-05, OPP-04, gave clear and reproducible results and were thus used on all samples, whereas

OPB04, OPT04, OPP06 were not taken in account because of differences between replicates were too

great. OPB-02, OPB-05, OPB-06, OPB-13, OPB-14, OPB-16, OPT-02, OPT-03, OPP-01, OPP-02,

OPP-03, OPP-05 were tested but did not amplify and OPB-03 showed only one monomorphic band.

The results for this primer were thus discarded.

Data analysis

RAPD markers were scored in a binary form as presence or absence of amplified bands

(respectively 1 and 0) for each sample.

Cluster analysis

We performed cluster analyses using the CLUSTER package

(http://www.biology.ualberta.ca/jbrzusto) to determine if samples of the same species formed groups

according to their morphological appearance. The asymmetrical Jaccard's coefficient, which considered

71


only shared presence, was used to calculate similarity between samples and to generate a similarity

matrix. This latter was then used to produce an UPGMA (Unweighted Pair-Group with Arithmetic

averaging) dendrogram, visualised with TREEVIEW (Page 1996)

Principal Coordinates Analysis

The similarity matrix calculated with Jaccard’s similarity coefficient between RAPD’s samples

was converted into a distance matrix (D=1-S). The distance matrix was used to perform Principal

Coordinates Analysis (PCoA ) using the R4 (Beta version) package (P. Casgrain & P. Legendre,

Université de Montréal). Results were graphically represented in a bivariate Scattergram using StatView

(SAS institute, Inc.)

In order to test the statistical significance of groups determined by both cluster and PCoA

analyses, a Mantel test was performed (999 permutations) using the R4 (Beta version) package. The

similarity matrix obtained by genetic data (excluding the outgroup species O. gunilla serdolis and H.

zarepha flexibilis), and a generated matrix in which a distance value of 1 was assigned between two

samples of a same subspecies and a value of 0 was assigned between two samples of different

subspecies, were converted into a distance matrix and compared pairwise.

Analysis of Molecular Variance (AMOVA)

Analysis of Molecular Variance (AMOVA; Excoffier et al, 1992) was performed using the

software ARLEQUIN 1.1 (Schneider et al 1997) in order to describe the genetic variability among and

within subspecies and populations. Only the data for the two Oleria onega subspecies were considered

in these analyses, avoiding hybrids and the outgroup. One population of O. o. agarista, that of

Ahuashiyacu, represented by fewer than five individuals was not taken into account.

72


RESULTS

RAPD markers

A total of 92 fragments were scored, the size of which ranged between 300 bp and 1500 bp. All

the fragments were polymorphic among individuals.

Four of the total of 92 fragments were present only in the outgroup species. Thus, out of the remaining

88 fragment present in the Oleria onega complex only, 57 were present in both subspecies, 76 were

present in O. o. ssp and 69 in O. o. agarista. Out of the 57 fragments common to O. o. ssp and O. o.

agarista, 30 were present in the hybrids. Five fragments specific to O. o. ssp. and only two specific to O.

o. agarista were found in hybrids. In some cases, fragments that were scarce in the subspecies, were

frequent in the hybrids, but no fragment was specific to hybrids.

The primers used varied widely in their ability to detect variation between and within

populations. Indeed, the mean number of polymorphic fragments among populations, scored

individually for each primers varied between 1.89(OPP-04) and 8.55 (OPT-01) (Table 2)

Table 2 : Characteristics of fragment variation generated by seven oligonucleotide primers in the RAPD analysis of nine Oleria onega ssp. populations.

Operon primer code Total fragment polymorphic

Mean no. of polymorphic fragments per population

OPB01 10 02.89 OPB11 12 06.22 OPB12 14 04.44 OPB15 14 06.89 OPT01 18 08.55 OPT05 17 08.44 OPP04 6 01.89 Mean 11.44 Total of polymorphic fragment 91

73


Data analysis

Cluster analysis based on the RAPD-generated similarity matrix separated the individuals into

five main groups (Figure 1). Group I is constituted by the different populations of O. o ssp. However,

while most of the individuals of the populations of San Roque (SR), Shapaja (Sh) and Shilcayo (S) are

gathered in this group, those from Ahuashiyacu and most of those from Urahuasha show a higher level

of within-population genetic diversity and are indeed dispersed throughout the dendrogram. Group II is

composed of the O. o. agarista individuals from population Km30 (K) and the two individuals of O. o.

agarista of the Ahuashiyacu mixed population (A49, A51). Hybrids from Shapaja (HSH) and the O. o.

agarista individuals from the Km28 (KM) population, as well as some individuals of O. o. ssp. from

Urahuasha (U76-U78) are clustered together in Group III. Finally, the groups IV and V include,

correspond to the two species used as "outgroup", Oleria gunilla serdolis (SRogs) and Hyposcada

zarepha (SRhz and SHhz).

The results obtained by the Principal Coordinates Analysis showed that the first three principal

factors accounted for 12%, 10%, and 7% respectively, of the total variation (Figure 2). Despite these

relatively low values, the two Oleria onega subspecies and their hybrids are clearly separated (Figure

2). However, as was observed with the cluster analysis, some of the individuals of each subspecies are

not grouped according to their morphological appearance. Here again, hybrids are intermixed with the

O. o. agarista individuals of population Km28, which are geographically the most distant one population

Shapaja.

A weak but significant correlation was observed between the genetic data and the

morphological appearance of the Oleria onega individuals. Indeed, the Mantel correlation value

between the RAPDs based distance matrix and the generated matrix (where a distance value 1 was

assigned between individuals of a different subspecies and 0 between individuals of the same

subspecies) was r = 0.14 (p = 0.001).

74


SRhz131SHhz105SHhz106SRogs123SRogs124SRogs125SRogs126SRogs128SRogs129KM99KM102KM101KM103U78U76U77HSH120KM100HSH116HSH118HSH119HSH117HSH121HSH122K39K10K33A51K25K26K1K35K41A49K31K38K29K32K34K11K37K36K27K28U79S20A55A63A66A59S7S16A57S42S9S15S19S45S2S21A54S43S23S22S46SH115SH108SH111SH110SH114SH109SH112SH113SR84SR97SR98A73A67S8U80SR89SH107SR86A70U82A71A72SR85SR88SR94SR95SR90SR93SR92U81SR91

V IV IV III III II II I I

Figure 1 UPGMA dendrogram based on Jaccard similarity coefficient for Oleria onega ssp., O. o. agari

ta,

Hyposcada zarepha flexibilis.

Figure 1 UPGMA dendrogram based on Jaccard similarity coefficient for Oleria onega ssp., O. o. agari

ta,

Hyposcada zarepha flexibilis.

sta, O.

gunilla serdolis, and Hyposcada zarepha flexibilis. Group I is consituted by O. o. ssp., group II by O. o. agaris

group III by hybrids and one O. o. agarista population, group IV by O. gunilla serdolis and group V by

sta, O.

gunilla serdolis, and Hyposcada zarepha flexibilis. Group I is consituted by O. o. ssp., group II by O. o. agaris

group III by hybrids and one O. o. agarista population, group IV by O. gunilla serdolis and group V by

75


Figure 2 : ce

atrix on the total (92) RAPD markers.

MOVA analysis

Partitioning of molecular variance was calculated among subspecies and among populations

ing subspecies as a variable group, 22.26% of the total variability was attributable to

differen

Principal components analysis of 87 Oleria onega butterflies based on a squared Jaccard distan

-.5

-.4

-.3

-.2

-.1

0

.1

.2

.3

Fact

or 2

-.5 -.4 -.3 -.2 -.1 0 .1 .2 .3 .4

Urahuasha, O.o.spp

Shilcayo, O.o.spp

Shapaja, O.o.spp

Shapaja, Hybrids

San Roque, O.o.spp

Km30, O.o.agarista

Km28, O.o.agarista

Ahuashiyacu, O.o.spp

Ahuashiyacu, O.o.agarista

Factor 1Factor 1 (12 %)

Fact

or 2

(10

%)

m

A

(Table 3). When us

ces among subspecies and 77.74% to differences within subspecies. Grouping the material

according to Oleria onega geographical populations, only 13.22% of the variability occurred between

subspecies, 19.59% among populations within subspecies, and 67.20% within subspecies. The pvalues

for all analyses of variance were highly significant (Table 3).

76


Table 3 : AMOVA for RAPD phenotypes in Oleria onega agarista and O. o. ssp. The total dataset contains 80

individuals from 7 populations (5 O. o. ssp. and 2 O. o. agarista), using 92 RAPD markers. Two analyses were

onducted: the first among and within the two subspecies and their hybrids, and the second among subspecies and

n

component dices

c

their populations and within populations. Percentile distribution of the variance components, as well as P value

and fixation indices are given. F statitstics are defined by three fixation indices : Fct=proportion of differentiatio

between subspecies, Fsc=differentiation among populations within subspecies, Fst=the global differentiation of

populations.

Source of Variation d.f. Variance % total variance p-value Fixation in

Among subspecies 2 2.59 Va 22.26 p<0.001 FST 0.2226

Within subspecies 84 77.74 p<0.001

9.04 Vb

Among subspecies .58 Va 3.22 <0.001

mong populations .25 Vb 9.59 <0.001 SC 0.2257 ithin subspecies ST 0.3280

CT 0.1321

1 1 1 p A 2 2 1 p FW F F Within populations 71 7.71 Vc 67.20 p<0.001 Significance test (1023 p rmutations)

DISCUSSION

Genetic Markers

The large set of markers obtained in this study confirmed the ability of the RAPD technique to

isms at the genetic level. We detected a high level of polymorphism between the two

studied

e

differentiate organ

subspecies, but also within each subspecies. Indeed, almost none of the amplified fragments

were present in all the individuals. Surprisingly, the high polymorphism of the studied organisms led to

difficulty in interpretation of the results. Indeed, only few markers were constantly present within

subspecies and even populations. It was thus difficult to assess whether these fragments were specific

to a subspecies or simply more frequent in some populations than in others.

77


Since RAPD markers are known to be highly polymorphic, it would be interesting to use more

conserved markers (i.e. isozymes) or to investigate more conserved sequences, in order to better

assess

Differentiation of the subspecies and hybrids

Both Cluster and PCoA analyses separated O. o. ssp and O. o. agarista populations into two

en them, lower than that separating both taxa from O.

gunilla

e two studied subspecies in a genetic analysis (e.g.

Guadag

Hybrids occurrence and viability

In the Ahuashiyacu population where we found both Oleria onega subspecies, no morphological

apaja hybrids are frequent. In addition, the genetic distance

betwee

the genetic relationships between these butterflies.

distinct groups. The relative distance betwe

serdolis, support their classification as two different subspecies (Figure 2). However, the

differentiation of the studied individuals in two groups (i.e. subspecies) was not clear, as demonstrated

by the low correlation value (Mantel's r = 0.14, P = 0.001) between RAPD markers based matrix and the

built distance matrices. This low value can be explained by the high polymorphism detected within the

studied subspecies, which partly hides the differences between them, as demonstrated by the AMOVA

results. In addition, some of the individuals of O. o. ssp. (U76, 77, 78) were grouped close to O. o.

agarista both by cluster and PCoA analyses and the population Km28 was closer to the hybrids than to

the other populations of the same species.

Independently of the type of marker used, one would expect the F1 hybrids between two taxa to

have a position intermediate between th

nuolo et al. 2001b). In the present study, the morphologically detected hybrids seem to be more

closely related to O. o. agarista than to O. o. ssp (Figure 2). The degree of hybridisation and the

probability of backcrossing with either or both parents were impossible to calculate as too few hybrids

were collected, and because of the high polymorphism among individuals and populations.

Nonetheless, our results suggest that at least some of the "hybrids" could be the result a backrossing

between a F1 hybrid and O. o. agarista as the recurrent parent.

hybrids were observed, whereas in Sh

n the Ahuashiyacu O. o. ssp and O. o. agarista individuals is lower than that between Shapaja

O. o. ssp and hybrid individuals. It is difficult to speculate on the reasons for these results, but they

78


allow us to formulate several hypotheses. The first is that a reproductive barrier exist between the two

subspecies in the population of Ahuashiyacu, that could lead to a genetic isolation. Unfortunately, it was

impossible to determine whether such a barrier reducing hybridisation exists. Indeed, neither the two

subspecies nor the hybrids ever reproduced in captivity, and it was thus impossible to perform crosses

in laboratory. Nevertheless, natural morphological hybrids from Shapaja did not show reduced viability,

and produced fertilised eggs with normal development. These observations suggest that genetic

incompatibility between O. o. ssp. and O. o. agarista, if it exists, is only partial. A second hypothesis is

that barriers to gene flow are not just associated with a few strongly selected colour pattern loci, but

dispersed across the genome as found on the sister group Heliconiinae (Jiggins et al. 1997), and are a

result of divergence in mate preferences, warning colour and ecology without hybrid inviability or sterility

(Jiggins et al., 1996 McMillan et al, 1997). Even so, a decisive explanation for the occurrence of

hybridisation in one zone of sympatry and not in the other, is hard to give.

Differentiation and relatedness between populations

Because of the recent modification of the environment by extensive deforestation, with the

bability of current gene flow between populations

is low.

f the Km28 population is difficult to explain, since Urahuasha is the upper population on the

SW slope (fig 1, p. 19) and is separated from KM28 by the highest peaks of the Cerro Escalera.

consequent fragmentation of natural habitats, the pro

The Cerro Escalera represents an additional physical barrier, which is thought to completely

separate the populations of Estero from those of Tarapoto. However, our results suggest, that gene

exchange between populations, as well as between Oleria onega subspecies, has occurred at least in

the past. The results of the AMOVA, showing a higher genetic variation within populations rather than

between them, is a pattern frequently observed in both plants and animals (Skotnicki et al., 1999;

Comes & Abbott, 2000; Moya et al., 2001). In the present case, this is an additional indicator of a high

rate of gene flow. The high percentage of polymorphic markers that was observed within subspecies,

becomes drastically lower at the population level (Table 3). In reality, this is due to the fact that the

majority of fragments were absent in one or other population but considered as present at the

subspecies level. Only a low percentage (max. 15 %) of monomorphic fragments was observed within a

population.

The similarity of some O. o. ssp individuals of Urahuasha (U76, 77, 78) with the O. o. agarista

individuals o

79


Moreov

ith the hybrids, while the two O. o. agarista individuals from the SW

e two studied subspecies. Subsequent

ters in common with O. o. agarista than with O. o. ssp.

Howeve

er, only O. o. ssp. individuals were found in Urahuasha and only O. o. agarista at Km28. The

possibility of a recent contact between these populations is thus extremely unlikely. Moreover, the other

Urahuasha individuals are mixed with the Shapaja and San Roque populations, which are the

geographically most distant ones. The most probable explanation we can offer for this fact, is that the

Urahuasha population has been in contact with O. o. agarista in the past, and that these three

individuals may be a results of hybridisation and subsequent back-crosses between F1 hybrids and O.

o. ssp. as the recurrent parents.

Another unexpected result is the clear separation between the two O. o. agarista populations of the NE

slope of the Cerro Escalera (Km30 and Km28), by both cluster and PcoA results. Moreover, the Km28

population was always grouped w

slope (Ahuashiyacu) were grouped with the Km30 population.

Since the Km28 population of O. o. agarista is not only genetically close to the analysed hybrids but

also relatively well separated from the other populations of the same species, it is conceivable that

these are the results of past hybridisation events between th

isolation from O. o. ssp. in Km28, and from O. o. agarista in Urahuasha could have led to a convergent

evolution. That would mean that both subspecies are under the effect of a speciation process, which

could lead to their extinction through hybridisation, as already shown by several studies on both plant

and animals (Huxel, 1999; Wolf et al. 2001).

It should also be noted that the hybrids from Shapaja, are more closely related to the three O. o.

ssp. individuals from Urahuasha cited above (U76, 77, 78) than to those from Shapaja. Morphologically

(data not shown), hybrids have more charac

r, we found them to share five RAPD characters in common with O. o. ssp. versus two with O.

o. agarista. This may suggest that the morphological characters of O. o. agarista are dominant over

those of O. o. ssp., but examination of a larger number of hybrids may be required to make this

assessment, as well as studies of genes coding for the morphology. On the other hand, hybridisation

may lead to the introgression of only neutral alleles, leaving intact parts of the genome that are under

selection and define species identity (Buerkle et al., 2000); this is translated by organisms that show

hybrid ancestry but retain the pure phenotype. Moreover, because of selection against recombinant

phenotypes, morphological measurement may consistently underestimate the true proportion of hybrids

in true phenotypes (Rieseberg et al., 1999a, 1999b)

80


Another important observation is that, during the three years of field work, the number of O. o.

agarista individuals of Shapaja and Ahuashiyacu, were decreasing rapidly (data not shown). Therefore,

at least in Shapaja, a selective pressure seems to act against this species, while the hybrids possess

some e

CONCLUSION

The objective of this study was to analyse genetic variation among two Oleria onega subspecies

e hybrids. We tried also to understand the speciation process and variation among

opulations in relation to their geographical situation. The most pertinent finding of this study was that

the two

cological or physiological advantages that allow them to persist in their environment. In another

study, it has been shown that extinction process may occur in five or fewer generations (Wolf et al.,

2001), a result that fits with our study. In order to test this hypothesis, it would be interesting to

genetically resample the Ahuashiyacu population, where no morphological hybrids were found, after

several generations. Nevertheless, the similarity of all the O. o. agarista individuals of the Km28

population with the morphological hybrids could be an additional indication that support this hypothesis,

since hybrid phenotype in Km28 may have disappeared even though gene information is still present.

As discussed above, this population could be the result of past hybridisation events, and where

selective pressure has acted in favour of the O. o. agarista form. From our results, we can postulate that

populations near the top of the mountain may have been in contact in the past, but that selective

pressures led them to adapt their morphological features to the pattern corresponding to the area: that

of O. o. agarista on the NE slope of the mountain and that of O. o. ssp. on the SW slope.

and their relativ

p

subspecies are distinct, even though polymorphism among population is high within each

subspecies, and that they are well distinct from the outgroup. Variation is greater within populations

than among them, nevertheless we believe that gene flow between them has occurred in the past (and

still occur when allowed by topography and climate). Hybrids are distinct from both subspecies, but the

population near the top of the mountain still shows traits of probable hybridisation events in the past. On

both slopes, selective pressures had led the rare forms to evolve toward the more common one, and

the only population where hybrids seem to suffer less evolution contraints is Estero, near Shapaja.

81


In this study, the choice of RAPD markers were helpful in detecting the differences between

subspecies, in identifying hybrids and indescribing variation within and among populations. RAPD

ACKNOWLEDGMENTS

pported by a grant of the Commission for travel’s of the Swiss Society of

atural Science (SANW), and the Matthey-Wüthrich funds for travel. Thanks are due to Pr. Ph. Küpfer

(Univ. O

REFERENCES

l, L.; Descimon, H.; Michel, F. 1999. Molecular phylogeny of swallowtail butterflies of

e tribe Papilionini (Papilionidea, Lepidoptera). Molecular Phylogenetics and Evolution 12, No2: 156-

en exotic ( Spartina alterniflora) and native (S. foliosa) cordgrass (Poaceae) in California, USA

amplification provided a large set of polymorphic markers. This is not astonishing as this technique can

detect variation in both coding and non-coding sequences, and the length of the primers allows the

amplification of a large number of fragments with a single primer. Nevertheless, further studies with

other markers are required to confirm the taxonomical statute of each subspecies, and to measure the

extend of hybridisation between them. It would be interesting to study the loci coding for the colour-

pattern differences and determine whether one of the patterns is dominant over the other.

This work was su

N

f Neuchâtel) for the use of material in his laboratory, to Dr. G. Lamas (Universidad Mayor de

San Marcos, Lima) for determination of Ithomiinae, to Dr K. Gotthard (University of Neuchâtel), Dr.

Niklas Wahlberg (University of Stockholm), and Dr. Laurent Amsellem and Dr. Michael Ròstas

(University of Neuchâtel) for helpful comments on the manuscript.

Aubert J.; Lega

th

167.

Ayres, D.R.; Garcia-Rossi, D.; Davis, H. G. ; Strong, D. R. 1999. Extent and degree of hybridization

betwe

determined by random amplified polymorphic DNA ( RAPD). Molecular Ecology 8: 1179- 1186.

Bartish, I.V.; Jeppson, N.; Nybom, H. 1999. Population genetic structure in the dioecious pioneer

82


plant species Hippophae rhamnoides investigated by random amplified polymorphic DNA (RAPD)

-503.

; Asmussen, M.A.; Rieseberg, L.H. 1999. The likelihood of homoploid

ae). Molecular ecology 8: 775-789.

) along European coasts.

raphical break in the Mediterranean ragwort Senecio gallicus Vill.

on to human mitochondrial DNA restriction data. Genetics

Philosophical Transactions of the Royal Society, London, Series B 353: 219-230.

tivum L.) to

of

barley (Hordeum marinum s.str. Huds) and bearded

elt, and

icrosatellites. Genome 44: 1-12.

markers. Molecular Ecology 8: 791-802.

Barton, N.H.; Hewitt, G.M. 1985. Analysis of hybrid zones. Annual Review of Ecology and

Systematics. 16: 113-148.

Barton, N.H.; Hewitt, G.M. 1989. Adaptation, speciation and hybrid zones. Nature 341: 497

Buerkle, C.A.; Morris, R.J.

hybrid speciation. Heredity 84: 441-451.

Bussell, J.D. 1999. The distribution of random amplified polymorphic DNA (RAPD) diversity amongst

population of Isotoma petraea (Lobeliace

Clausing, G.; Vickers, K.; Kadereit, J.W. 2000. Historical biogeography in a linear system : genetic

variation of Sea Rocket (Cakile maritima) and Sea Holly (Eryngium maritimum

Molecular Ecology 9: 1823 . 1833.

Comes, H.P.; Abbott, R.J. 2000. Random amplified polymorphi DNA (RAPD) and quantitative trait

analyses across a major phylogeog

(Asteraceae). Molecular Ecology 9: 61-76.

Excoffier, L.; Smouse, P.E.; Quattro, J.M. 1992. Analysis of molecular variance inferred from metric

distances among DNA haplotypes: applicati

131: 479-491.

Gaston, K.J. 1998. Species-range size distributions: products of speciation, extinction and

transformation.

Guadagnuolo, R.; Savova-Bianchi, D.; Felber. 2001a. Gene flow from wheat (Triticum eas

jointed goatgrass (Aegilops cylindrica Host.), as revealed by RAPD and microsatellite markers.

Theorical and Applied Genetics 103: 1-8.

Guadagnuolo, R.; Savova-Bianchi, D.; Felber. F. 2001b. Search for evidence of introgression

wheat (Triticum eastivum L.) traits into sea

wheatgrass (Elymus caninus L.) in central and northaern Europe, using isozymes, RAPD and

microsatellite markers. Theorical and Applied Genetics in press.

Guadagnuolo, R.; Savova-Bianchi, D.; Felber. 2001c. Specific genetic markers for wheat, sp

four wild relatives: comparison of isozymes, RAPD's, and wheat m

Hewitt, G.M. 1988. Hybrid zones- natural laboratories for evolutionary studies. TREE 3: 158-167.

83


Huff, D.R.; Peakall, R.; Smouse, P.E. 1993. RAPD variation within and among natural populations of

6:

onservation 89: 143-152.

to and H. himera (Lepidoptera: Nymphalidae). Biological Journal

dity, 79, 495

du Languedoc. Montpellier: Académie de

e of

tic forms of the butterfly Heliconius numata. Evolutionary Ecology 13: 721-754.

:

d zones. In: Harrisson, R.G. ed. Hybrid Zones and the Evolutionary process. New-York:

.

of the National Academy of Sciences of the United States of America 94:

ci (Hemiptera: Aleyrodidae) based on random amplified polymorphic DNA-polymerase

.

Introgression between cultivated sunflowers and a

outcrossing buffalograss, (Buchloë dactyloides (Nutt.) Engelm). Theoretical and Applied Genetics 8

927-934.

Huxel, G.R. 1999. Rapid displacement of native species by invasive species : effects of hybridization.

Biological C

Jiggins, C.D.; McMillan, W.O.; Neukirchen, W.; Mallet, J. 1996. What can hybrid zones tell us about

speciation? The case of Heliconius era

of the Linnean. Society 59: 221-242.

Jiggins, C. D.; McMillan, W.O.; King, P.; Mallet, J.,1997. The maintenance of species differences

across a Heliconius hybrid zone, Here

Joron, M. 2000. Coloration avertissante et mimetisme müllérien: le problème de la diversification Thése

de Doctorat. Université des Sciences et Technique

Montpellier.

Joron, M.; Wynn, I.R.; Lamas, G.; Mallet, J.L.B. 2001. Variable selection and the coexistenc

multiple mime

Mallet, J.L.B. 1989. Analysis of clines and linkage disequilibria in Heliconius butterflies. Heredity 62

283-284.

Mallet, J.L.B. 1993. Speciation raciation and color pattern evolution in Heliconius butterflies; evidence

from hybri

Oxford University Press. pp. 226-260.

Mallet, J.L.B.; Barton, N. 1989. Inference from clines stabilized by frequency-dependent selection

Genetics 122: 967-976.

McMillan, W.O.; Jiggins, C.D.; Mallet, J.L.B. 1997. What initiates speciation in passion-vine

butterflies? Proceedings

8628-8633.

Moya, A.; Goya, P.; Cifuentes, F.; Cenis, J.L. 2001. Genetic diversity of Iberian populations of

Bemisia taba

chain reaction. Molecular Ecology 10: 891-897.

Page, R.D.M. 1996. TREEVIEW: An application to display phylogenetic trees on personal computers

Rieseberg, L.H.; Kim, M.J.; Seiler, G.J. 1999a.

84


sympatric wild relative, Helianthus petiolaris (Asteraceae). International Journal of plant Sciences 160:

102-108.

Rieseberg, L.H.; Whiton, J.; Gardner, K. 1999b. Hybrid zones and the genetic architecture of a barrie

to gene flo

r

w between two wild sunflower species. Genetics 152: 713-727.Applications in the

nfirm a

phical variation in the bushcricket Ephippiger ephippiger (Orthoptera:

e for

niversity of Geneva.

glaciale on Ross Island, East Antartica. Molecular Ecology 8: 753-762.

Molecular

isms

primers are useful as genetic markers. Nucleic Acids Research 18: 6531-6535.

Biosciences 12: 357-358.

Ritchie, M.; Kidd, D.M.; Gleason, J.M. 2001. Mitochondrial DNA variation and GIS analysis co

secondary origin of geogra

Tettigoniideae), and resurrect two subspecies. Molecular Ecology 10: 603-611.

Schneider, S.; Kuefer, J.M.; Roessli, D.; Excoffier, L. 1997. ARLEQUIN Version 1.1: A softwar

Population Genetic Data Analysis Genetics and Biometry Laboratory. Geneva: U


Waiblingen.

Skotnicki, M.L.; Ninham, J.A.; Selkirk, P.M. 1999. Genetic diversity and dispersal of the moss

Sarconeurum

Vucetich, L.M., Vucetich, J.A.; Joshi, C.P.; Waite, T.A.; Peterson R.O. 2001. Genetic (RAPD)

diversity in Peromyscus maniculatus populations in a naturally fragmented landscape.

Ecology 10: 35 - 40.

Wiesing, K.; Nybom, H.; Wolff, K.; Meyer, W. 1995. DNA fingerprinting in plants and fungi. CRC

Press, Boca Raton.

Williams, J.G.K.; Kubelik A.R.; Livak K.J., Rafalski J.A.; Tingey S.V. 1990. DNA polymorph

amplified by arbitrary

Wolf, D.; Takebayashi, N.; Rieseberg, L. H. 2001. Predicting the risk of extinction through

hybridization. Conservation Biology 15, No 4: 1039-1053.

85

Chapter 4

Modelling of the Distribution of Oleria onega agarista

and Oleria onega ssp. (Ithomiinae, Lepidoptera), in

Peru Using Geographic Information Systems and

Geospatial Analysis

Based on:

Gallusser, S.; Bouzelboudjen, M.; Rahier, M.

Modelling of the Distribution of Oleria onega agarista and Oleria onega ssp. (Ithomiinae,

Lepidoptera), in Peru Using Geographic Information Systems and Geospatial Analysis.

Unpublished chapter

Chapter 4 Modelling of the distribution of two Ithomiinae subspecies

INTRODUCTION

The Oleria onega subspecies complex

Oleria onega is a complex of ithomiinea butterfly subspecies (Lamas, pers comm.) that extends

to all the south-American tropical forest and counts some fifteen subspecies, some of them are yet

undescribed. Two of these subspecies are found near Tarapoto, in north eastern Peru: O. o. agarista

(C. Felder and R. Felder) and O .o. ssp. a recently discovered subspecies. Their taxonomical status is

not completely defined, between subspecies or two totally different subspecies (G. Lamas, pers. com.).

Their genetical variations (Chapter 3) revealed that gene flow had occurred between populations (and

may still occur when allowed by topography and landuse). These two subspecies are geographically

separated by a mountain chain, the “Cerro Ecalera” which constitutes a big biological barrier, with a

quite different fauna and flora on each side (Mallet, 1989, 1993; Shulte, 1999; Joron 2000, 2001). O. o.

ssp. lives mostly on the southwestern (SW) slope, whereas O. o. agarista extend his range on the

northeastern (NE) slope and all the Huallaga Valley till Yurimaguas. Four Solanaceae species are used

by Oleria onega as host plants ( Mallet pers. comm.) : Solanum mite (Ruiz & Pav.), Solanum anceps

(Ruiz & Pav.), Solanum angustialatum (Bitter), and Solanum uleanum (Bitter) (Knapp 1997). On the NE

slope, three of them are found together: S. mite, S. anceps, and S. uleanum; whereas on the SW slope

only S. mite was found. S. angustialatum grows on the upper part of the mountain.

Geographical information systems and their recent applications in biology

In the strictest sense, a GIS is a computer system capable of assembling, storing, manipulating,

and displaying geographically referenced information, i.e. data identified according to their locations.

GIS is not only a cartographic computation system, but also an analytical tool that allows storage,

manipulation and analysis of data. GIS makes the link between discrete and continues variables by a

complex tissue of spatial relations on the same map, and allows treating quantitative and qualitative

data. Several applications have been used in environmental sciences. For example, the soil structure

and functioning (Cosandey et al., 2002) have been studied in 2D and 3D (Mendonca et al., 2000).

86


The application of GIS to biological sciences increased in the last decade with a focus on

habitat and species distribution on different organisms indeed birds (Maurer, 1994), plants of the Rocky

Mountains (Morain et al, 1993), and aquatic plants (Lehmann, et al, 1994). The different applications of

GIS in studies of animal communities were widely described by Mc Laren and Braun (1993).

Epidemiological applications were also studied with Lyme disease (Nicholson et al., 1996). In

entomology, GIS has been used for study of population structure and geographical variation, using a

combination of pure GIS and other spatial geostatistic programs such as Surfer and Isatis (Cesaroni et

al, 1997; Kidd and Ritchie, 2000; Margraf, unpublished). A parallel between geographical and genetical

(RAPD's) variation was recently demonstrated in Orthoptera (Kidd and Ritchie, 2000; Ritchie et al,

2001).

Using GIS to examine the Oleria onega subspecies complex

The two sides of the Cerro Escalera are climatically different: the NE slope, benefits from a

cooler and wetter climate due to clouds coming from the Atlantic that are retained by the mountain,

whereas the SW is hotter and dryer. The hotter climate on the SW slope is accentuated by high

deforestation because of the proximity of the city, and the resulting open areas constitute a factor

restricting the range of the Oleria populations. Study of molecular variation revealed that populations

are well distinct, but that gene flow may occurs between them (Chapter 3), due to possible contacts in

the past, before the increasing deforestation. Even though the Cerro Escalera is a topographical barrier,

Oleria onega agarista probably found a dispersal path around the mountain because we observed two

mixed populations: Ahuashiyacu where the two subspecies co-occur, and Shapaja (not represented in

this study) where they hybridise.

Data on distribution of butterflies and their host plants, as well as environmental data, for factors

such as temperature, humidity, topography and landuse, were analysed In twenty-three sites. The

presence and absence of butterflies, and of the different host plant species, the type of soil, the

dimension of the trunk and the density of the understorey were noted. Thereafter, temperature, humidity

and the butterfly density were measured at least two times per month. Through statistical and

interpolation methods, both the spatial distribution of Oleria onega subspecies and environmental data

were calculated by kriging and conditional simulations. Thirty other sites were introduced to obtain

results concordant with reality, because most of the study area is inaccessible due to topography. For

87


these synthetic sites no observations on vegetation and soil were available. However, climatic

parameters and the butterfly density were estimated based on the data of the most similar study sites.

Therefore the interest of this work is in the modelling aspect and the way to use mapping and statistical

tools in biology rather than an assertion about actual patterns in the field.

MATERIAL AND METHODS

Ecological database on O. onega ssp. and O. o. agarista and environmental data:

The creation of a spatial (Arc/Info, ArcView; ESRI, 1996, 1997) and tabular database (Excel;

Microsoft, 1997) represented the first stage of work. This database integrated ecological data relevant

for the study and is subdivided into two complementary parts; on the one hand, a spatial database

incorporating various coverages (a set of thematically associated data considered as a unit, e.g. soils,

streams, roads, or land use) (ESRI, 1996). This spatial database contains discrete data (geographic

features containing boundaries: point, line or area boundaries) and continuous data (topography,

temperature, humidity). On the other hand a descriptive database including data relating to the Oleria

species (23 stations) was built. Before the implementing of this database, validation procedures

adapted for each type of data were defined and applied in a systematic way. To facilitate its data

management, a simple structure (tabular files) was adopted for the descriptive database.

The number of males and females observed per day were counted in twenty-three study sites

between Tarapoto and the Pongo de Cainarachi, from September to February 1999-2000 and 2000-

2001. Data were collected from two to twelve times per month for each site. The coordinates of the sites

were established with a Garmin GPS. Each site represents an area of approx. 100x 100 m. As Oleria

butterflies are active mostly in the morning (Chapter 1 and 2) the number of individuals per day, is in

reality the number of individuals observed in the morning (between 8.00h-12.00 h). In each site, the

presence or absence and the species of the host plant were noted, as well as soil type, the forest type,

and understorey density. Temperature and humidity were measured at least two times per month.

The sites near the road were easily accessible, but not sufficient to allow data interpolation.

Therefore 30 synthetic data sites were added, with estimations for mean number of males and females,

humidity and temperature, according to the data in the most similar observation site.

88


Data storage and map construction

A simplified map was drawn from Geographical map No 1658, scale of 1:100,000 (Instituto

Nacional Geográfico del Peru). In this simplified map (1cm = 425 m), the two main rivers (Shilcayo, and

Progreso river), the road to Yurimaguas (Carretera Marginal), twenty altitudinal contour lines with

intervals of 200m, and the limit of the study area were represented as well as the twenty-three

observation sites. The limits of areas under different landuse were visually assessed in the field and

drawn on the map. All these features were digitalized and corrected using ArcInfo. Lines (rivers, roads,

isolines of topography) and polygons (landuse and study area) were generated, as well as a 1x1cm grid

including all the study area. Thirty synthetic points were added (from Id 24 to 53, see annex 1), pointing

out the chosen place on the View and introducing them as a theme. To each point (observed +

synthetic) correspond mean values for number of female and of male, mean relative humidity, and

mean temperature degree. The means were done per month, but also the mean for all the different

data. The Annex 1 summarize all data used for the geospatial analysis. These represent the mean of

observations from September 2000 to February 2001.

Geospatial analysis: geostatistics, kriging, simulations and probabilities

The stochastic approach, based on kriging and multiple conditional simulations, leads to a

quantification of uncertainty for both Oleria spatial distribution and environmental data.

To calculate the distribution of Oleria onega spp. and Oleria onega agarista and environmental

data (topography, temperature, humidity, etc.) we used a probabilistic approach developed by Matheron

[1970]. In fact, direct knowledge of all these parameters in the field can be only partial. The term

"regionalized" was proposed by Matheron (1970) to qualify a phenomenon developing in space (and\or

in the time) and showing a certain structure. We treated regionalized variables by using the probabilistic

theory of the random functions and interprets the regionalized variable as a "realization of the random

function" (Chilès & Delfiner 1999). The work of interrogation of the data, then the modelling of their

structural properties, establishes under the name of variographic analysis the inescapable phase of any

concrete geostatistical study [Chauvet, 1992]. In the probabilistic models, the simplest tool to measure

an estimator of quality is the variance. One makes the hypothesis that for a vector h, the increase Z(x +

h) - Z(x) has an expected value and a variance independent at the point x, it is the intrinsic hypothesis :

89


E Z(x+h)−Z(x)[ ]=0 (1)

Var Z(x+h)−Z(x)[ ]=2γ(h) (2)

The function γ(h) is called semi-variogram. The semi-variogram is calculated by:

γ(h)= 12N(h) (z(xi+h)−z(xi))2[ ]

i=1

N(h)∑

(3)

z(xi) are the data, xi the points for which the data are available at the same moment in xi and xi + h and

N(h) is the number of couples of points distant from h. In any geostatistical study the hypotheses of

stationarity and ergodicity should be verified to allow correct estimations (Chauvet, 1992). The Strict

Stationarity is a particular case of great pratical importance is when the finite-dimensional distributions

are invariant under an arbitrary translation (Chilès&Delfiner (1999). Such Random Functions is called

stationarity. Physically this means that the phenomenon is homogeneous in space and, so to speak,

repeats itself in the whole space. The sand in the jar is a good image of a stationarity random function in

three dimensions, at least if the sand is well sorted (otherwise, if the jar vibrates, the finer grains will

eventually seep to the bottom, creating nonstationarity in the vertical dimension). The Ergodicity is an

intimidating concept. The practioner has heard that the Random Functions should be ergodic, since

“this is what makes statistical inference possible,” but is no sure how to check this fact and proceeds

anyway, feeling vaguely guilty of having perhaps overlooked something very important. We will attempt

here to clarify the issues. In practice, ergodicity is never a problem. When no replication is possible, as

with purely spatial phenomenon, we can safety choose an ergodic model. If the phenomenon is

repeatable, typically time-dependant fields or simulations, averages are computed over the different

realisations, and the only issue (more a physical than a mathematical one) is to make sure that we are

not mixing essentially different functions.

Having fitted a model to the experimental variogram , an estimation of our variable can be made

at all points of a regular grid by the method of standard kriging (Geovariances, 1998). There are other

estimation methods taking into account non-stationarity of the phenomenon (universal kriging, Intrinsic

Random Function of order k : IRF-k), (Chauvet, 1992).

Simulations represent a method of estimation which reconstruct the real variability of the

variable and allow calculation of probability maps (Chilès and Delfiner, 1999). If Z(x) is the studied

regionalised variable, which we consider as a realisation of a IRF-k, and which is known only at n points

x, one calls conditional simulation of Z(x) any realisation T(x) of the same IRF-k that passes by the data

90


at points x. A conditional simulation is thus characterised by the following properties : it has the same

generalised covariance as the phenomenon considered, it passes through the experimental points and

has a similar distribution (histogram). Although the conditional simulations are not good estimators of

the real field (the best estimator being the kriging) they are possible variants of the real field presenting

the same degree of variability, respecting what one knows of the real field. They allow to surround the

possible fluctuations in the phenomenon (Kimmeier and al., 2001) better than do kriging (which

smoothes reality) and the standard deviation of kriging (which gives no indication of the spatial structure

of the error of estimation). They remain however connected to kriging : if one builds a great number of

conditional simulations, the average of their values at a point will restore the estimated value by kriging,

and their variance of corresponding estimation.

The files with the 53 points with their co-ordinates and mean number of females, males, mean

relative humidity, and mean temperature (calculated from September to February) were treated with the

Isatis software (Geovariances, 1998). An exploratory analysis of the spatial features of the data was

done, then the generalised covariance for each variable was calculated, as well as the variogram fitting.

Then a cross validation was done for each variable to build the kriging and its standard deviation. From

the Cross validation, conditional simulation and probability maps were calculated. Kriging, simulations

and probabilities were created on a grid of 0.5 X 0.5 cm.

Interpolation and data grid analyses

Kriging, probabilities and simulations from Isatis were interpolated for the density of females and

males, humidity and temperature, with a cell size of 0.5 x0.5 cm, on ArcView. Using the grid-analyst, the

maps were extracted from the grid into new maps corresponding to the study area, to study the possible

relations between the different variables.

Buffers were created on study sites, rivers and road, using Arcview. For each element, the

extension for the buffers was done according to the estimation of their possible interactions with the

environment. The buffers of the sites were 200, 400, and 600m, those of the road 100 and 200m and

those of the rivers 100, 200 and 300 m. Two buffers were done on the rivers, with the gradient scale of

the probabilities to find two males and two females.

91


Relations between the variables (i.e., topography, humidity, temperature, density of females and

males, probabilities of densities of females and males, and buffers on road, rivers and study sites) were

assessed visually, and interpreted based to field observation.

RESULTS

Cartography :

In the cartographic part, the twenty-three study sites and the thirty synthetic sites were first

represented (fig 1b) on the study area. Five maps were then created with the different numerised

themes. On each map, the road, the rivers, the contour line, the sites (only observation sites), the grid,

and the limit of the study zone (polygon + line themes) were represented. On these five maps, the

distribution of plants and animal communities, and soils information were represented for each study

site. As no data were collected on the synthetic sites, they are not represented on these maps.

• presence or absence of the different butterfly subspecies (fig 2a): on the SW slope of the

mountain, O. o. ssp.. was present in only two sites, and we found a third site (site 7) where the

two butterfly subspecies were sympatric. On the NE slope (from the site 15) only O. o. agarista

was present. These results assess our hypothesis that the Cerro Escalera is a biological barrier

for our organisms, even though some dispersal pathways are possible, as demonstrated by the

sympatry of the two subspecies in site 7.

• host plant species (fig 2b) : on the SW slope, only S. mite was collected, whereas on the NE

side S. mite, S. anceps, and S. uleanum were found on the mountain, but S. anceps was

inexistent in the valley of Progreso. The only host plant that grew on the upper part of the

mountain was S. angustialatum. Therefore S. anceps and S. angustialatum seem to be more

frequent at high elevation than in the valley. Another pattern that may be related to altitude was

noticed with S. uleanum and S. mite: the plants tend to be dark red with increased pilosity in site

18 (alt 800m) whereas the individuals of the valley were green.

• forest type (fig 2c) : trees with a diameter greater than 70 cm were found in the less deforested

zone, mostly on the NE slope, whereas on the SW slope, trees were smaller and cultivated area

mare extend more extended. Butterflies were found only in the forested sites.

92


• understorey abundance (fig 2d) :. in the forested sites, density of the understorey was slightly

higher on the NE slope where forest were less disturbed.

• soil type (fig 2e) : the soil was mostly brown (with abundance of clay) with some areas or yellow

sand in the valleys. In the forested areas, the percentage of humus was obviously higher. On

the mountain, some areas were constituted by white sand and humus, which limited the

diversity of plants, butterflies and other organisms. On the upper part, bare rock appears,

partially covered by a humus cover up to one meter thick depending on the topography. This

suggests that the absence of butterflies in these sites may be due not only the effect of altitude,

but also to the soil quality that affect the presence of the host plants and their quality.

On the map of topography (fig 3) we can observe the Cerro Escalera that divides the study area,

with slopes of more than 60 degrees (fig 4). The highest peaks reach an altitude of some 1400 m. The

SW versant extends to the Cumbaza Valley (Tarapoto) at 300m alt. The NE versant revealed a rougher

topography, divided by the small Valley of Progreso which extends to the Amazon Basin.

Landuse shows deforested areas near Tarapoto, and some cultures near Progreso. All the

mountainous areas were still forested areas. Some small open areas, not represented on the map, do

exist.

Geostatistic: Kriging, Simulations and probabilities maps

Results of the exploratory and variographic analysis of the spatial features of the females,

males, humidity and temperature data as well as the cross validations were not represented here.

Kriging and conditional simulations were based on the values of the fifty-three study sites.

Kriging was used to study the eventual relations between the density of butterflies, temperature, and

humidity, with the topography and landuse.

Simulations were conducted by the method of Turning bands (1000 bands) and were used as

graphical representation of the spatial variation of butterfly distributions on the study area. Four

simulation maps were done for males and four for females: one based on the dispersion, one on the

mean number of individuals, then the largest and the smallest number. The map of the mean of

individuals was similar to the kriging map (data not shown). From the simulations, five probability maps

were calculated for the males and females according to the mean number of individuals observed per

day on an area of ca. 100x100m. This mean number varied between 0 and 5, therefore probabilities

93


were calculated for 1, 2, 3,4, and 5 individuals (annex 2b). The coloured scale from green to red,

revealed the probability (%) of observing this mean number of butterflies on the different points of the

grill. For males and females, the probability of observing one individual was high, whereas the

probability of observing five individuals was too low to be concordant with reality (data not shown). The

sites of highest density were the areas near Shilcayo (site 3) and Km30 (site 18) (the two green areas

on the map).

Interpolation and data grid analyse

The data from Isatis were imported to ArcView and visualization was effected on the different

variables of the three Isatis results, Kriging, Simulations and Probabilities, and were extracted to

correspond to the study area. In order to examine correlation between butterfly distribution and

environmental variables, only the kriging and one of the probabilities maps were used and represented

here. The probability of finding 2 individuals was the representation that best fit with the reality (fig. 10-

11).

The kriging of temperature (fig. 6a and b) revealed that near the top of the mountain and near

rivers temperatures were slightly lower (24-26°), and increased to 29° C near the town and in

deforested areas. Standard deviations of temperature were low near the sites of measurement, but

increased drastically away from the site. Mean temperature predominant in the study area was 26-27°

C. In the kriging of humidity (fig. 7a and b), the highest means were in the Km30 area (till 95%),

whereas in the Tarapoto area means are situated between 78-83%. The river edge (more obvious with

the Progreso river) is clearly represented by a mean of 88-90%. Standard deviations of humidity

increased slightly with the distance from the measurement points.

The kriging results of densities of males and females per day were fairly similar (fig. 8a and b,

9a and b) with two high densities sites (km30 and Shilcayo). On the NE slope as well as the

Ahuashiyacu area, density was uniform with means between 0.7-1.5 ind. Standard deviations of density

of males showed greater uniformity and increased less than those of the females.

The most representative probability was that with two individuals per day (fig 10, 11). The probability of

finding two males per day was higher than for females on the NE slope, whereas the highest values of

probabilities of finding two females were concentrated in the Km30 site and Shilcayo.

94


Relationships between butterfly distribution and environmental data

Comparing visually all the kriging and probabilities maps with the topography and landuse grid,

and the three buffers, the following relationships are apparent:

- the forested areas show means for temperature between 26-27°C and means for humidity of 85-90%.

Forested areas predominate up to 500 m. alt. and deforestation is greater on the SW slope (near the

town), then on the NE slope.

- For the kriging of density of both males and females, values between 0.8-1.5 ind./per day were

predominant in all the area and these values were found mostly between 85-90% of humidity and 25-

27° C. Butterflies were found between 300-800m and in forested areas.

- The probability of finding two ind./day was the closest to reality, and this probability was higher in

forested area, mostly on the NE slope. The probability decreased with altitude. Males were easier to

find in the lowland than females.

- The highest concentration of butterflies occurred in the Shilcayo and Km30 areas. Both sites are on

the mountain flanks but with moderate slopes. On both sites, soil type is similar with clay and with high

humus content; on the other sites on the mountain flanks, slopes are more abrupt and rock and white

sand predominated.

- According to the four kriging maps, landuse and the topography, a plausible hypothesis for the

presence of O. o. agarista in Ahuashiyacu, on the SW slope, is that they found a pathway passing at the

side limit of mountain, where temperature and humidity are optimal for the butterflies. As no topographic

data were available to join the Shilcayo site with the Km30, and as O. o. agarista was never found in

other parts of the SW slope, this is the only hypothesis we can suggest. This will be discussed further.

- The probability of finding two individuals in relation with the presence of the rivers is higher at the

origin of the two rivers (fig 13 and 14). Probabilities were higher near the Progreso river, which flows in

areas less deforested than does the Shilcayo river. However we can observe that the proximity of the

road reduce the probabilities for the Progreso river (this pattern is more marked for females). Higher

temperatures near the road may explain this. No relationships were found between the buffers, the

kriging maps and the probability maps.

Distribution of the butterflies is dictated first by food availability. Host plant distribution depends

on soil quality, environment (forest) and altitude. Plants were found between 300 and 800m, always in

forested areas. A second factor that allows the presence of butterflies is climate (temperature and

95


humidity). We found that the more suitable parameters for the butterflies were between 25-27° C and

85-90% humidity. These two factors are obviously dependent on the vegetation, and on altitude. In this

area, the extend of deforestation depends on topography. The mountains are less deforested because

of their steep slopes and poor soil quality. However, temperature and humidity varied greatly between

sunny and rainy days. Butterflies were easier to find just after rain, or when the weather was cloudy. In

these cases their density increased up to 15 - 20 per morning, whereas during the hottest days, they

stay in the vegetation, moving as little as possible (probably early in the morning and at dawn), and are

hardly observed. This lack of action during the day explains that the highest probabilities are of finding

only two individuals per morning in sites where they are probably quite abundant.

With map-based data there is a strong tendency to ignore or not recognise issues of data

quality ( Matthews, 1990). Spatial data are subject to the to the same errors and limitations to which any

compiled database is subjected. By their very nature, spatial data have an associated spatial accuracy,

that is, how close to reality are the shapes and distances depicted on the map. For example, spatial

errors can occur because of the resolution of the data. GIS technology allows researchers to combine

data from widely disparate scales, projections, or levels of precision. However, one must consider the

reliability of the data. For example, a study conducted on Lyme disease considered the abundance of

ticks in their environment and the prevalence of Borrelia burgdorferi infection in ticks through

comparison of kriging, allowing predictions of Lyme disease risk in humans (Nicholson & Mather, 1996).

GIS and geospatial analysis may be extended to speciation events, and can be very useful in predicting

movements of populations and species. Recent studies conducted on a hybrid zone, demonstrated the

correlation between Mitochondrial DNA variation and geographical variations in bushcrickets (Ritchie et

al, 2001) and the patterns and causes of geographic variation among their populations according to

their environment (Kidd & Ricthie, 2000). In our study area, most of the species of warningly coloured

butterflies and poisonous frogs are known to meet and hybridise with sibling species or with other

morphological phenotype of the same species (Mallet, 1989, 1993; Shulte 1999; Joron 2000, 2001), and

it could be interesting to examine structure of relationships between mimicry rings, and environmental

variations by a geostatistical approach.

96


Fig 1a : Study area location (Tarapoto, Peru)

97


98


99


100


101


102


CONCLUSION AND REMAINING QUESTIONS

The injection of synthetic data allowed us to obtain good interpolations of the different variables.

The probabilities of the presence of butterflies, as well as a kriging of temperature and humidity were

possible to estimate in inaccessible regions. However, we do not know how well these results fit. The

results showed that the butterflies inhabit an altitude range between 300 and 800m. Their presence is

correlated with temperature and humidity, regulated by the forest, but depends most of all on the

presence of host plants. The Cerro Escalera constitutes a barrier not only for the butterflies but also for

the host plants. The presence of the host plant species is correlated with soil quality and altitude.

Through the kriging of temperature and humidity, we can observe that interactions between the butterfly

populations on opposite slopes of the mountain may be possible. However the absence of the host

plant on the upper part of the mountain may make such dispersal imposible.

Further studies should be done with more field observations and more data on the different

sites, allowing a better establishment of the means of the different variables. Geographical distance

should be correlated with genetical distances and estimation of the gene flow between populations,

allowing a better understanding of dispersal pathways over or around the mountain.

Even so, this study remains a demonstration of the great possibilities to study the interactions

between the environmental variables and organisms through a set of GIS and geospatial analysis

techniques.

REFERENCES

Cesaroni, D.; Matarazzo, P.; Allegrucci, G.; Sbordoni, V. 1997. Comparing patterns of geographic

variation in cave crickets by combining geostatistic methods and Mantel tests. Journal of Biogeogrpahy

24: 419-431.

Chauvet, P. 1992. Traitement des données à support spatial : la géostatistique et ses usages. 43 p.

Fontainebleau, Paris, France: Centre de Géostatistique, Ecole des Mines de Paris.

Chilès J.P.; Delfiner, P. 1999. Geostatistics : modelling spatial uncertainty. Probability and

Mathematical Statistics, Wiley Series. 695 p.

103


Cosandey, A.C.; Guenat, C.; Bouzelboudjen, M.; Maître, V.; Bovier, R. 2002. The modeling of

functional units of soil process based on a three-dimensional soil-horizon cartography: an example

applied to denitrification in riparian zone. Geoderma (in press).

ESRI. 1996. ArcView GIS, The Geographic Information System for Everyone. California, USA.

Environmental Systems Research Institute.

ESRI. 1997. Understanding GIS. The Arc/Info Method. Version 7 for Unix and Open VMS. ed. New

York.: Environmental Systems Research Institute, California, USA. Wiley & Sons,.

Geovariances. 1998. Isatis 3, The geostatistical key. User's guide. Avon, France.

Golden Software 1999. User's guide. Surfer For Windows, version 7.: Golden software, Inc.

Joron, M. 2000. Coloration avertissante et mimetisme müllérien: le problème de la diversification Thèse

de Doctorat. Université des Sciences et Technique du Languedoc. Montpellier: Académie de

Montpellier.



Kidd, D.M.; Ritchie, M.G. 2000. Inferring the patterns and causes of geographic variation in Ephippiger

ephippiger (Orthoptera, Tettigoniidae) using geographical information systems (GIS). Biological Journal

of the Linnean Society 71: 269-295.

Kimmeier, F.; Bouzelboudjen, M.; Ababou, R.; Ribeiro, L. 2001. Travel path uncertainty : a case

study combining stochastic and deterministic hydrodynamic models in the Rhône valley, Switzerland.

3rd International Conference on Future Groundwater Resources At Risk. 25-27 June 2001 - Lisbon,

Portugal, 1-14.

Knapp, S.; Helgason, T. 1997. A revision of Solanum section Pteroidea: Solanaceae. Bull. nat. Hist.

Mus. London 27, no1: 31 - 73.

Lehmann, A.; Jacquet, J.M.; Lachavanne, J. B. 1994. Contribution to a GIS to submerged

macrophyte biomass estimation and community structure modeling, Lake Geneva, Switzerland. Aquatic

Botany 58: 347-361.

Mallet, J. L.B. 1989. The genetics of warning color in Peruvian hybrid zones of Heliconius erato and

Heliconius melpomene. Proc. R. Soc. Lond. B. Biol. Sci. 236: 163-185.

Mallet, J. L.B. 1993. Speciation raciation and color pattern evolution in Heliconius butterflies evidence

from hybrid zones. In: Harrison RG, ed. Hybrid Zones and the Evolutionary Process. New-York, USA:

104


Oxford University Press. 226-260.

Margraf, N. 1998. Oreina elongata (Suffr.) (Coleoptera: Chrysomelidae) Répartition spatiale en fonction

de différents facteurs écologiques. Analyse au travers d'une cartographie GIS. LEAE. Neuchâtel:

Université de Neuchâtel,.pp. 108. Unpublished Diploma thesis.

Matheron, G. 1970. La théorie des variables régionalisées et ses applications. Centre de

Géostatistique, Ecole des Mines de Paris, Paris.

Matthews, S.A. 1990. Epidemiology using a GIS : the need for caution. Comput. Environ. Urban. Sys

14: 213-221.

Maurer, B. 1994. Geographical population analysis: Tolls for the analysis of biodiversity. Blackwell

scientific publications. Oxford.

McLaren, S. B.; Braun, J. K. 1993. GIS application in Mammalogy Special publication of the Oklahoma

Museum of Natural History, in colaboration with the Carnegie Museum of Natural History. Oklahoma. 1-

41.

Mendonça, S. M. L.; Guenat, C.; Bouzelboudjen, M.; Golay, F. 2000. Three-dimensional GIS

cartography applied to the study of the spatial variation of soil horizons in a Swiss floodplain. Geoderma

97: 351-366.

Morain, S.A.; Neville, P.R.H.; Budge, T.K.; Morrison, S.C.; Helfrich, D.A.; Fritch, S. 1993. Design

and test of an object oriented GIS to map plant species on the southern Rockies. Geocarto International

4: 33-44.

Nicholson, M. C.; Mather, T.N. 1996. Methods for evaluating Lyme disease risks using geographical

information system and geospatial analysis. Enotmological Society of America, Forum 33 (5): 711-720.

Ritchie, M.; Kidd, D. M.; Gleason, J.M. 2001. Mitochondrial DNA variation and GIS analysis confirm a

secondary origin of geographical variation in the bushcricket Ephippiger ephippiger (Orthoptera:

Tettigoniideae), and resurrect two subspecies. Molecular Ecology 10: 603-611.


Waiblingen.

105

ANNEXES

Annexes

Annex 1 : Datas for interpolations

ID X Y Mean T Mean H Mean F Mean M 001 -01.510 04.894 026.00 091.30 000.00 000.00 002 -00.723 07.513 027.40 087.40 001.00 001.00 003 00.824 12.809 026.80 089.00 003.30 003.40 004 02.637 05.399 029.50 079.30 000.00 000.00 005 03.397 03.041 032.80 073.60 001.00 001.00 006 05.163 03.400 028.50 081.00 000.00 000.00 007 08.442 05.444 027.50 084.50 001.88 001.24 008 06.873 05.547 031.30 074.00 000.00 000.00 009 06.245 08.884 031.30 074.00 000.00 000.00 010 04.653 08.874 026.00 088.30 001.10 001.40 011 10.666 11.536 025.00 085.50 000.00 000.00 012 16.412 10.730 025.00 087.70 000.00 000.00 013 14.772 13.111 025.10 091.60 000.00 000.00 014 12.139 16.180 026.10 089.70 000.00 000.00 015 09.104 20.799 025.50 090.90 000.90 001.60 016 08.333 22.778 027.30 087.50 000.00 000.00 017 06.970 22.701 021.50 099.00 002.50 003.00 018 06.233 24.091 023.00 096.50 004.30 001.50 019 18.341 17.171 027.00 090.00 000.20 001.20 020 23.881 17.423 026.00 090.00 001.20 001.00 021 24.035 21.195 026.80 088.80 000.20 000.50 022 20.307 27.152 025.50 090.00 000.40 000.20 023 16.807 33.601 026.70 086.30 001.00 000.70 024 00.120 02.610 026.00 091.33 000.00 000.00 025 00.650 05.120 032.75 073.60 000.00 000.00 026 03.430 00.750 032.75 073.60 000.00 000.00 027 07.850 01.200 032.75 073.60 000.00 000.00 028 00.320 08.240 027.42 087.44 001.00 001.00 029 01.170 09.300 028.50 080.96 000.00 000.00 030 04.610 07.500 028.50 080.96 000.00 000.00 031 10.610 03.720 028.50 080.96 000.00 000.00 032 09.370 06.870 026.00 088.29 001.10 001.40 033 09.770 05.200 026.00 088.29 001.10 001.40 034 02.360 10.200 026.00 088.29 001.10 001.40 035 -00.660 14.270 026.00 088.00 003.30 003.40 036 00.830 14.720 026.00 088.00 003.30 003.40 037 04.130 12.790 026.00 088.00 003.30 003.40 038 13.600 09.460 025.00 087.70 000.00 000.00 039 15.990 15.860 025.00 087.70 000.00 000.00 040 08.390 18.640 026.07 089.70 000.00 000.00 041 16.990 21.710 027.00 090.00 000.20 001.20 042 20.620 24.250 027.00 090.00 000.20 001.20 043 24.820 23.350 027.00 090.00 000.20 001.20 044 19.290 25.890 027.00 090.00 000.20 001.20 045 15.090 30.410 026.67 086.33 001.00 000.70 046 12.680 34.170 026.67 086.33 001.00 000.70 047 04.130 24.120 023.00 096.50 004.30 001.50 048 02.170 21.050 025.08 091.60 000.00 000.00 049 20.040 14.350 025.50 092.00 000.20 001.20 050 24.540 13.910 025.50 092.00 000.20 001.20 051 23.610 10.670 025.50 092.00 000.20 001.20 052 17.240 14.290 025.50 092.00 000.20 001.20 053 12.020 19.780 025.50 092.00 000.20 001.20 study sites Id 1 to Id 23 and synthetic sites id 24 to 53

106

Annexes

ANNEX 2a : Kriging and kriging standard deviation of the mean number of males. (cell size 0.5x0.5 cm)

ANNEX 2b : Probabilities of males present per day from September to February (cell size 0.5x0.5 cm)

Probability of 1 male per day Probability of 2 male per day

107

Annexes

Probability of 3 male per day Probability of 4 male per day

Annex 3a : Quick statistics on data : Correlation Matrix (total number of samples = 53)

Variable

Temp_Moy (°)

Hum_percent

MoyF

MoyM

Temp_Moy (°)

1.00

-0.94

-0.38

-0.37

Hum_percent

-0.94

1.00

0.35

0.41

MoyF

-0.38

0.35

1.00

0.79

MoyM

-0.37

0.41

0.79

1.00

Moy = Mean, F = Female, M = Male, Temp = Temperature (°) Hum_ percent = Humidity (%)

Annex 3b : Quick statistics on Kriging Variable Nb samples Minimum Maximum Mean Std.

Dev. Variance Skewness Kurtosis

T_krig 4340 24.00 30.02 26.84 0.50 0.25 1.89 15.32 T_stdv 4340 1.79 2.10 2.05 0.08 0.01 -1.65 4.61 H_krig 4340 77.14 95.69 87.35 2.22 4.91 -1.63 8.81 H_stdv 4340 4.44 6.09 5.66 0.50 0.25 -0.83 2.31 F_krig 4340 -0.00 3.69 0.72 0.56 0.31 2.38 9.64 F_stdv 4340 0.68 1.06 0.91 0.12 0.01 -0.17 1.56 M_krig 4340 0.08 3.12 0.82 0.43 0.19 1.97 9.82 M_stdv 4340 0.66 1.00 0.87 0.11 0.01 -0.26 1.60 krig = kriging, stdv = standard deviation, T = Temperature, H = Humidity, F = Female, M = Male,

Annex 3c : Quick statistics on Kriging

Variable Nb samples

Minimum Maximum Mean Std. Dev. Variance Skewness Kurtosis

Simu_MoyF_Me 4340 0.12 3.31 0.86 0.33 0.11 2.41 12.49 Simu_MoyF_D 4340 0.96 2.81 1.81 0.28 0.08 0.21 2.98 Simu_MoyF_L 4340 2.55 8.30 4.26 0.68 0.46 0.73 4.11 Simu_MoyF_S 4340 -5.38 0.80 -2.50 0.66 0.43 0.08 4.33 Simu_MoyM_Me 4340 -0.04 3.74 0.70 0.57 0.32 2.31 9.33 Simu_MoyM_D 4340 0.18 1.76 0.78 0.31 0.09 0.09 1.92 Simu_MoyM_L 4340 0.95 5.87 2.90 0.82 0.67 0.35 2.78 Simu_MoyM_S 4340 -3.84 2.46 -1.49 0.79 0.62 1.14 6.83 Simu = simulation, Moy = mean, F = Female, M = Male, Me = mean, D = display, S = smallest, L = largest.

108

CONCLUSION and OUTLOOK

Conclusion and Outlook

CONCLUSION AND OUTLOOK

1. Are the four Solanum species host plants of Oleria onega agarista and O. o. ssp.? How does

oviposition behaviour of the two Oleria subspecies differ? The most important result of the

experiments conducted in the cages is that Solanum mite is the most preferred host plant of both

subspecies. Switches to other Solanum species occur when S. mite is rare or absent. O. o. ssp.

which lives on the SW slope of the mountain, lays most of its eggs on objects close to the plant,

whereas O. o. agarista, which lives on the NE slope, laid them on the plant itself. A possible

explanation may be that the risk of predation on eggs when laid on the host plant is higher on the

SW slope and that this selected for a new oviposition strategy. The cost of oviposition away from

the plant may be compensated by a higher survival rate of eggs and by the higher host plant

density found on the SW side of the mountain

2. Is oviposition preference correlated with preference and performance of the larvae? Is there a

difference in potential predation pressure on larvae of both subspecies, that may affect their

behaviour? Is there differences in potential predators species for different oviposition

environments? Among the four host plants tested, Solanum mite was the most preferred host by

the larvae and preference among species was highly correlated with larval performance. Larvae

reared on S. mite were bigger and survival was greater on this plant than on S. anceps.

Therefore female and larval choices were highly correlated with each other and with larval

performance.

Comparing the larval behaviour of both Oleria subspecies, we found that O. o. agarista larvae

move more than the these of O. o. ssp. larvae. This difference may be related to the fact that this

subspecies suffers less predation than the other, but also that S. mite is less abundant on the NE

slope and that larvae would have to wander more to find the host plants if laid on other

substrates. When eggs were artificially deposited in the field, the eggs placed on the SW slope

showed a higher survivorship when laid on other objects than on the host plant. When host plant

stems, were protected by sticky glue in the SW slope, egg disappearance was significantly higher

on non- protected plants. Taken together, these observations led to the hypotheses that O. o.

ssp. might suffer more predation in its narural environment than does O. o. agarista, and

109


therefore that predation was higher on the SW slope. In the field, 70-80% of the potential

predators collected on both slopes of the mountain were ants. Among them, the only abundant

genus of predators that was found only on the SW slope and not on the NE side was the

Ponerinea genus Ectatomma. We cannot assert that Ectatomma are the ultimate cause of

oviposition behaviour of O. o.ssp., as no acts of predation were observed when eggs were

offered to the ants. However, we observed that Ectatomma individuals spend most of their time

on plants (Solanum spp. and others), that not only are they present on the SW slope only, but

also abundant there, and that they are considered as potential predators in the literature.

Therefore we suggest that they may constitute a selective pressure, either because of predation

on eggs, or because they might repel ovipositing females, leading to the behaviour of laying eggs

next to the host plant.

3. Are the two Oleria subspecies genetically distinct ? The results showed that the two subspecies

are distinct and that together they are well separated from the outgroup. Hybrids are intermediate

between subspecies, but more closely related to some populations near the top of the mountain.

These populations still show traits of probable hybridisation events in the past. Genetic variation

was higher among populations than between them. Nevertheless, we believe that gene flow had

occurred between them in the recent past (max. ten years ago). On both slopes, selective

pressures had led the rare forms to evolve to the more common pattern, and the only population

where hybrids seem to suffer less constraint is Estero, near Shapaja.

4. Is there a relationships between distribution of Oleria and environmental factors such as altitude,

temperature, humidity and other factors? The results have shown that the butterflies inhabit an

altitude range between 300 and 800m. Their presence is correlated with temperature and

humidity, regulated by the forest cover, but depends most of all on the presence of host plants.

The abiotic conditions in which the butterflies were the most abundant were at temperatures

between 26-27°C and relative humidity between 85-90%. The Cerro Escalera constitutes a

barrier not only for the butterflies but also for the host plants. The presence of the different host

plant species is correlated with soil quality (sand, clay, rocks) and altitude. Through the kriging of

temperature and humidity, we can observe that interactions between butterfly populations on the

110


two slopes the mountain may be possible. However, the absence of host plants on the upper part

of the mountain make this unlikely. The use of synthetic data allowed us to conduct interpolations

of the different variables. The probabilities of butterfly presence, as well as kriging of temperature

and humidity then permitted estimation in inaccessible region. However we do not know how well

these results fit with reality, and they remain a hypothesis to be confirmed in the filed. Even so,

this study serves as a demonstration of the great opportunities to study the interactions between

the environmental variables and studied organisms through a set of GIS and geospatial analysis

techniques.

Based on the results presented, a set of new questions arose :

1. To define better the suitability of the four host plant for the larvae, it would be interesting to conduct

an analysis of their chemical compounds to determine whether they emit different volatiles that may

affect the female choice.

2. The relationships between ants and butterflies need further observations. Predation events on eggs

and larvae should be observed in the field as well as the possible influence of ants on ovipositing

females. The extrafloral nectaries appear not to occur in the studied Solanum species and the

relation Ithomiinae-Solanum-Ecatomma need further study. One case has already been cited in the

literature of possible mutualistic interactions between Ectatomma ants and Mechanitis isthmia

(Ithomiinae) (Young, 1978) in which the larvae provide exsudates to the ants when feeding on

Solanum sp.

3. Results obtained with molecular markers revealed that further studies with other markers are

required to assert the real taxonomical statute of the two subspecies, and to measure traits of

hybridisation degree between them. Further studies should include other Oleria species present in

the area (O. lerida lerida and O. gunilla serdolis), focusing also on the possible hybridization with

the two forms of O. lerida lerida.

4. Another interesting point with genetic studies is a better understanding of the loci coding for the

colour-pattern differences, for oviposition behaviour and differences between behavioural ability and

define if one of the pattern is dominant over the others.

111


5. The relations between butterflies and their environmental parameters requires further study with

more field observation and more data on the different sites, allowing a better establishment of the

means of the different variables. Correlation between geographical distance and genetical

distances should be examined, as well as gene flow between populations, allowing a better

understanding of the dispersal pathways that exist between populations on the two slopes of

mountain. Studies of Orthoptera have indicated that GIS and molecular results taken together

might help us to understand speciation processes (Kidd & Ritchie, 2000; Ritchie et al. 2001).

REFERENCES

Kidd, D.M.; Ritchie, M.G. 2000. Inferring the patterns and causes of geographic variation in Ephippiger

ephippiger (Orthoptera, Tettigoniidae) using geographical information systems (GIS). Biological Journal

of the Linnean Society 71: 269-295.

Ritchie, M.; Kidd, D.M.; Gleason, J.M. 2001. Mitochondrial DNA variation and GIS analysis confirm a

secondary origin of geographical variation in the bushcricket Ephippiger ephippiger (Orthoptera:

Tettigoniidae), and resurrect two subspecies. Molecular Ecology 10: 603-611.

Young, A.M. 1978. Possible evolution of mutualism between Mechanitis caterpillars and an ant in

northeastern Costa Rica. Biotropica 10 (1): 77-78.

112

CURRICULUM VITAE

Professional address Personal address

University of Neuchâtel rue de l’Université 20

Institute of Zoology, L.E.A.E. 1005 Lausanne, Switzerland

C.P.2, CH-2007 Neuchâtel tel: ++41-(0)21-320 52 01

Switzerland calle La Merced s/n, Tarapoto - Perú.

email: [email protected] Tel: ++51-(0)94-69 00 71

[email protected]

Date of Birth : 10 January 1973

Birthplace: Lausanne-Switzerland

Nationality : Swiss - Argentinian - Peruvian

Civil status : married

Languages in order of fluency : French (native language) Spanish (fluent spoken and written), English,

basic German

REFERENCE :

Prof. Martine Rahier: ++41 (0)32 718 31 37, University of Neuchâtel, Switzerland

[email protected]

EDUCATION :

1979-1988 Primary and secondary school at the "Ecole Rudolf Steiner", of Lausanne.

1988-1991 "prégymnasiale" and "gymnase" at the “Ecole Nouvelle de la Suisse Romande”,

in Lausanne.

113

mailto:[email protected]

mailto:[email protected]

1991 “ Baccalauréat scientifique ” (exams in Mathematics, Physics, Biology,

Chemistry, Geography, History, German, English, French).

1992-1996 Studies in Biology at the University of Neuchâtel, Switzerland.

oct. 1996 “Diplôme” in biology with specialisation in Entomology. Title of Diploma thesis : “

Etude des différences morphométriques entre populations d’Oreina cacaliae,

speciosissima et elongata (Chrysomelidae)” (Spatial morphological

differentiation of the three species) with Prof. Martine Rahier.

1997-2001 PhD thesis at the University of Neuchâtel.( Prof. Martine Rahier): Biology,

Behaviour and Taxonomy of two Ithomiinae butterflies: Oleria onega agarista

and Oleria onega ssp. " in collaboration with Dr. James Mallet, University

College, London, and Dr. Gerardo Lamas, Universidad Nacional Mayor de San

Marcos, Lima, Perú.

TRAVEL AND PRACTICAL EXPERIENCE:

1991-1992 Travel in South America : Argentina, Paraguay, Peru, Bolivia and Colombia.

1996-1997 Travel in South America including :

Visits :

-Parque nacional de Iguazú , Pcia Misiones, Argentina,

-Parque Nacional del Pantanal (frontier of Brasil, Paraguay and Bolivia).

-Conections and visit at the entomological collections of the Universidad del

Valle, Cali, Colombia.

Field work :

nov. 1996 -with the Fundación Moises Bertoni of Asunción, Paraguay.

apr. 1997 -with students in Iquitos (with the Universidad Nacional de la Amazonia

peruana),

Teaching:

114

115

1997-2001 Participation in and organisation of practical courses in Entomology at the

University of Neuchâtel, with second year students.

1997-2001 Participation and organisation of practical courses in Soil Zoology at the

University of Neuchâtel, with third year students.

May 2001 Field work with third year students in Haute-Provence with Prof. J-D. Gallandat

(University of Neuchâtel, Switzerland).

PARTICIPATION IN CONGRESSES:

April 1997 -Primer congreso de manejo del medio ambiente, Universidad del Valle, Cali,

Colombia.

March 1998 - "Zoologia et Botanica", Geneva, Switzerland.

22-26 Oct. 2000 - XLII Convención Nacional de Entomologia, Tarapoto, Perú, Oral presentation:

"Biologia y comportamiento de dos Ithóminos de la Amazonia peruana"

PUBLICATIONS:

- Oviposition behaviour and host-plant preferences of Oleria onega agarista and

Oleria onega ssp. in north-eastern Peru (Ithomiinae, Nymphalidae), In prep.

- Genetic (RAPD) diversity between Oleria onega agarista and Oleria onega

ssp. (Ithomiinae, Nymphalidae) in north-eastern Peru. In prep.

- Larval performance, and predation effect of ants on two Ithomiinae

(Nymphalidae) butterfly subspecies. In prep.

Biology, Behaviour and Taxonomy of two Oleria onega ... · Chapter 4 Modelling of the Distribution...

Documents

Transcript of Biology, Behaviour and Taxonomy of two Oleria onega ... · Chapter 4 Modelling of the Distribution...